From 02db6466bd0f2ba39507431ab16493ae2b9a2e38 Mon Sep 17 00:00:00 2001 From: Luke van der Hoeven Date: Mon, 17 Dec 2012 23:12:17 -0500 Subject: [PATCH 1/6] update to my info --- 404.html | 18 +++++++++--------- _layouts/layout.html | 14 +++++++------- _layouts/post.html | 12 +++++------- about.html | 16 +++------------- 4 files changed, 24 insertions(+), 36 deletions(-) diff --git a/404.html b/404.html index 3c83d294b..10966df89 100644 --- a/404.html +++ b/404.html @@ -4,11 +4,11 @@ - Zach Holman + Luke van der Hoeven - - + + @@ -19,7 +19,7 @@ - + @@ -27,21 +27,21 @@
- +
- Zach - Holman + Luke + van der Hoeven
- - + +
diff --git a/_layouts/layout.html b/_layouts/layout.html index af150d6ba..aed0d806f 100644 --- a/_layouts/layout.html +++ b/_layouts/layout.html @@ -7,7 +7,7 @@ {{ page.title }} - + @@ -19,7 +19,7 @@ - + @@ -27,21 +27,21 @@
- +
- Zach - Holman + Luke + van der Hoeven
- - + +
diff --git a/_layouts/post.html b/_layouts/post.html index 74c316d27..6bbf52b14 100644 --- a/_layouts/post.html +++ b/_layouts/post.html @@ -17,18 +17,16 @@

Discussion, links, and tweets

- - + +

- I'm a developer at GitHub. Follow me on Twitter; - you'll enjoy my tweets. I take care to carefully craft each one. Or at least aim to make - you giggle. Or offended. One of those two— I haven't decided which yet. + I'm a software engineer.

- - + +
diff --git a/about.html b/about.html index eb706e1e1..77510d768 100644 --- a/about.html +++ b/about.html @@ -7,21 +7,11 @@
-

Zach Holman

+

Luke van der Hoeven

-

You can find Zach on Twitter or send him an email.

+

You can find Lukas on Twitter

-

Zach works at GitHub. When he's not - doing that, he's blogging, hacking on - open source, and giving talks about building products, - growing startups, and using Ruby. He lives in San Francisco.

- -

If you have any questions or feedback, check out the issues on holman/feedback - on GitHub. It's sort of like an "Ask Me Anything" setup, except it uses a - nerdy social network and crams it into a bug tracker.

+

I <3 Engineering.

From 1fcadc33b055b0451196c220178a23ae1a2380ea Mon Sep 17 00:00:00 2001 From: Luke van der Hoeven Date: Tue, 18 Dec 2012 00:13:39 -0500 Subject: [PATCH 2/6] add posts --- _config.yml | 4 +- _posts/2010-09-01-back-again.html | 31 +++ _posts/2010-09-01-brand-new.html | 18 ++ _posts/2010-09-01-collaboration.html | 26 +++ ...10-09-01-cool-ruby-plugin-for-the-day.html | 16 ++ _posts/2010-09-01-css-nav.html | 28 +++ _posts/2010-09-01-details.html | 21 ++ _posts/2010-09-01-development-to-the-max.html | 15 ++ .../2010-09-01-dynamic-form-generation.html | 103 +++++++++ _posts/2010-09-01-ie-gotcha-3.html | 20 ++ _posts/2010-09-01-mallassi-project-story.html | 10 + _posts/2010-09-01-moving-and-archiving.html | 16 ++ _posts/2010-09-01-my-take.html | 31 +++ _posts/2010-09-01-new-designs.html | 22 ++ _posts/2010-09-01-passenger-3-coming.html | 25 +++ _posts/2010-09-01-peerdrum-live.html | 14 ++ ...0-09-01-podcasting-skip-filter-and-ha.html | 64 ++++++ _posts/2010-09-01-rails-3-is-coming.html | 13 ++ _posts/2010-09-01-rails-3-progression.html | 14 ++ _posts/2010-09-01-servers.html | 14 ++ _posts/2010-09-01-steam-for-mac.html | 20 ++ ...10-09-01-they-said-it-couldnt-be-done.html | 16 ++ ...0-09-01-this-will-only-make-things-wo.html | 21 ++ _posts/2010-09-01-typekit.html | 12 ++ _posts/2010-09-01-vim-thoughts.html | 21 ++ ...0-09-02-amazing-how-much-turnaround-s.html | 10 + ...0-09-02-eating-large-chunks-of-banana.html | 10 + ...0-09-02-i-have-so-much-data-to-export.html | 10 + ...0-09-02-in-other-news-redis-is-actual.html | 10 + ...0-09-02-note-to-self-never-allow-upda.html | 10 + ...0-09-02-rt-jasonmauer-ping-the-itunes.html | 10 + ...0-09-02-rt-wilshipley-itunes-we-wont-.html | 10 + _posts/2010-09-02-well-thats-done.html | 10 + ...0-09-02-why-does-everybody-and-everyt.html | 10 + ...0-09-02-wow-browsers-are-getting-real.html | 10 + _posts/2010-09-08-a-tale-of-two-asyncs.html | 27 +++ .../2010-09-10-because-i-told-you-before.html | 16 ++ _posts/2010-09-23-kthxbye-1-0-0.html | 23 ++ ...0-09-25-dont-stay-in-bed-unless-you-c.html | 11 + _posts/2010-09-25-site.html | 22 ++ _posts/2010-09-30-messing-with-the-tubes.html | 41 ++++ _posts/2010-09-30-month-in-music.html | 12 ++ .../2010-10-04-ie-gotcha-4-ie7-edition.html | 13 ++ .../2010-10-05-kthxbye-1-1-0-rails-ified.html | 14 ++ ...0-10-06-internet-stop-being-so-intern.html | 13 ++ .../2010-10-09-new-project-in-the-works.html | 12 ++ _posts/2010-10-25-what-i-have-decided.html | 10 + _posts/2010-10-27-rails-3-gotcha.html | 14 ++ _posts/2010-12-26-include-vs-extend-in.html | 10 + ...010-12-26-job-new-company-salescrunch.html | 24 +++ ...0-12-26-rock-stars-of-code-when-a-cul.html | 53 +++++ ...0-12-27-rock-stars-of-code-2-a-side-e.html | 12 ++ .../2010-12-31-make-it-not-as-readable.html | 11 + _posts/2011-01-25-three-weeks-of.html | 42 ++++ ...1-03-08-an-open-letter-from-our-vp-of.html | 22 ++ ...1-03-28-read-the-damn-code-rtdc-episo.html | 24 +++ ...02-the-challenges-of-working-remotely.html | 10 + _posts/2011-04-04-possessive-tendencies.html | 18 ++ _posts/2011-04-06-salesschool-nyu.html | 197 ++++++++++++++++++ ...1-04-08-a-new-thing-starting-at-sales.html | 10 + ...1-04-12-on-meetings-from-a-meeting-so.html | 10 + _posts/2011-04-27-easter.html | 18 ++ _posts/2011-05-10-lessons-learned.html | 27 +++ _posts/2011-05-10-possessive-tendencies.html | 20 ++ ...1-05-10-top-skype-owner-microsoft-twe.html | 10 + _posts/2011-05-13-testing-sux.html | 31 +++ ...1-06-02-better-testing-framework-is-b.html | 16 ++ ...2011-06-14-bettering-rails-pagination.html | 16 ++ _posts/2011-06-20-listening-last-week.html | 10 + ...1-07-13-mostly-false-ive-been-listeni.html | 10 + ...1-07-13-why-osx-lion-is-a-consumer-dr.html | 18 ++ _posts/2011-07-14-first-jquery-hack.html | 79 +++++++ _posts/2011-08-18-designing-for-change.html | 44 ++++ ...2011-09-13-ie-gotchas-ie7-and-z-index.html | 30 +++ ...05-ruby-and-rails-described-perfectly.html | 15 ++ _posts/2011-10-26-i-love-specs.html | 16 ++ .../2012-02-26-choosing-a-path-is-hard.html | 41 ++++ ...012-02-27-bjarne-stroustrup-c11-style.html | 17 ++ _posts/2012-05-03-very-helpful.html | 11 + _posts/2012-12-12-left.markdown | 27 --- atom.xml | 20 +- posterous.rb | 85 ++++++++ 82 files changed, 1888 insertions(+), 39 deletions(-) create mode 100644 _posts/2010-09-01-back-again.html create mode 100644 _posts/2010-09-01-brand-new.html create mode 100644 _posts/2010-09-01-collaboration.html create mode 100644 _posts/2010-09-01-cool-ruby-plugin-for-the-day.html create mode 100644 _posts/2010-09-01-css-nav.html create mode 100644 _posts/2010-09-01-details.html create mode 100644 _posts/2010-09-01-development-to-the-max.html create mode 100644 _posts/2010-09-01-dynamic-form-generation.html create mode 100644 _posts/2010-09-01-ie-gotcha-3.html create mode 100644 _posts/2010-09-01-mallassi-project-story.html create mode 100644 _posts/2010-09-01-moving-and-archiving.html create mode 100644 _posts/2010-09-01-my-take.html create mode 100644 _posts/2010-09-01-new-designs.html create mode 100644 _posts/2010-09-01-passenger-3-coming.html create mode 100644 _posts/2010-09-01-peerdrum-live.html create mode 100644 _posts/2010-09-01-podcasting-skip-filter-and-ha.html create mode 100644 _posts/2010-09-01-rails-3-is-coming.html create mode 100644 _posts/2010-09-01-rails-3-progression.html create mode 100644 _posts/2010-09-01-servers.html create mode 100644 _posts/2010-09-01-steam-for-mac.html create mode 100644 _posts/2010-09-01-they-said-it-couldnt-be-done.html create mode 100644 _posts/2010-09-01-this-will-only-make-things-wo.html create mode 100644 _posts/2010-09-01-typekit.html create mode 100644 _posts/2010-09-01-vim-thoughts.html create mode 100644 _posts/2010-09-02-amazing-how-much-turnaround-s.html create mode 100644 _posts/2010-09-02-eating-large-chunks-of-banana.html create mode 100644 _posts/2010-09-02-i-have-so-much-data-to-export.html create mode 100644 _posts/2010-09-02-in-other-news-redis-is-actual.html create mode 100644 _posts/2010-09-02-note-to-self-never-allow-upda.html create mode 100644 _posts/2010-09-02-rt-jasonmauer-ping-the-itunes.html create mode 100644 _posts/2010-09-02-rt-wilshipley-itunes-we-wont-.html create mode 100644 _posts/2010-09-02-well-thats-done.html create mode 100644 _posts/2010-09-02-why-does-everybody-and-everyt.html create mode 100644 _posts/2010-09-02-wow-browsers-are-getting-real.html create mode 100644 _posts/2010-09-08-a-tale-of-two-asyncs.html create mode 100644 _posts/2010-09-10-because-i-told-you-before.html create mode 100644 _posts/2010-09-23-kthxbye-1-0-0.html create mode 100644 _posts/2010-09-25-dont-stay-in-bed-unless-you-c.html create mode 100644 _posts/2010-09-25-site.html create mode 100644 _posts/2010-09-30-messing-with-the-tubes.html create mode 100644 _posts/2010-09-30-month-in-music.html create mode 100644 _posts/2010-10-04-ie-gotcha-4-ie7-edition.html create mode 100644 _posts/2010-10-05-kthxbye-1-1-0-rails-ified.html create mode 100644 _posts/2010-10-06-internet-stop-being-so-intern.html create mode 100644 _posts/2010-10-09-new-project-in-the-works.html create mode 100644 _posts/2010-10-25-what-i-have-decided.html create mode 100644 _posts/2010-10-27-rails-3-gotcha.html create mode 100644 _posts/2010-12-26-include-vs-extend-in.html create mode 100644 _posts/2010-12-26-job-new-company-salescrunch.html create mode 100644 _posts/2010-12-26-rock-stars-of-code-when-a-cul.html create mode 100644 _posts/2010-12-27-rock-stars-of-code-2-a-side-e.html create mode 100644 _posts/2010-12-31-make-it-not-as-readable.html create mode 100644 _posts/2011-01-25-three-weeks-of.html create mode 100644 _posts/2011-03-08-an-open-letter-from-our-vp-of.html create mode 100644 _posts/2011-03-28-read-the-damn-code-rtdc-episo.html create mode 100644 _posts/2011-04-02-the-challenges-of-working-remotely.html create mode 100644 _posts/2011-04-04-possessive-tendencies.html create mode 100644 _posts/2011-04-06-salesschool-nyu.html create mode 100644 _posts/2011-04-08-a-new-thing-starting-at-sales.html create mode 100644 _posts/2011-04-12-on-meetings-from-a-meeting-so.html create mode 100644 _posts/2011-04-27-easter.html create mode 100644 _posts/2011-05-10-lessons-learned.html create mode 100644 _posts/2011-05-10-possessive-tendencies.html create mode 100644 _posts/2011-05-10-top-skype-owner-microsoft-twe.html create mode 100644 _posts/2011-05-13-testing-sux.html create mode 100644 _posts/2011-06-02-better-testing-framework-is-b.html create mode 100644 _posts/2011-06-14-bettering-rails-pagination.html create mode 100644 _posts/2011-06-20-listening-last-week.html create mode 100644 _posts/2011-07-13-mostly-false-ive-been-listeni.html create mode 100644 _posts/2011-07-13-why-osx-lion-is-a-consumer-dr.html create mode 100644 _posts/2011-07-14-first-jquery-hack.html create mode 100644 _posts/2011-08-18-designing-for-change.html create mode 100644 _posts/2011-09-13-ie-gotchas-ie7-and-z-index.html create mode 100644 _posts/2011-10-05-ruby-and-rails-described-perfectly.html create mode 100644 _posts/2011-10-26-i-love-specs.html create mode 100644 _posts/2012-02-26-choosing-a-path-is-hard.html create mode 100644 _posts/2012-02-27-bjarne-stroustrup-c11-style.html create mode 100644 _posts/2012-05-03-very-helpful.html delete mode 100644 _posts/2012-12-12-left.markdown create mode 100644 posterous.rb diff --git a/_config.yml b/_config.yml index 2a2a3ebbc..ea1810377 100644 --- a/_config.yml +++ b/_config.yml @@ -1,6 +1,6 @@ -markdown: rdiscount +markdown: redcarpet pygments: true permalink: /posts/:title auto: true -rdiscount: +redcarpet: extensions: [smart] diff --git a/_posts/2010-09-01-back-again.html b/_posts/2010-09-01-back-again.html new file mode 100644 index 000000000..dfd9af71e --- /dev/null +++ b/_posts/2010-09-01-back-again.html @@ -0,0 +1,31 @@ +--- +layout: post +title: Back again. +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047905729/back-again +posterous_slug: post/1047905729/back-again +--- +

Couple of posts upcoming.

+ +

First one will be on a mini-app I’m working on for Hargray. Its just kinda a shout-out to a cool little helper for Textile forms. Hargray peeps needed a quick way to do email blasts, since the Oracle Forms stuff doesn’t give a slick/easy way to format emails. So I got the request to build a simple form for emailing stuff to customers.

+ +

Well I’ve become a big fan of Textile. I use it right here for formatting my blog entries right here. Its just a super slick way to format text without the craziness of HTML markup. Plus its a little more intuitive for non-technical users. At first I figured, hey, I’ll just point email creators at the textile formatting guide and they can do all the markup by hand. But lo and behold, I found felttippens Textile-Editor-Helper. It adds a nifty little editor bar with a lot of the markup tools you’d expect to see so that you can simply press the button and it generates the markup for you.

+ +

Then I wanted to give the users a way to preview the actual email content before sending, so I created a little helper that does the following

+ +

+

  $("#render_preview").click(function (e) {

+

    e.preventDefault();

+

    $('.preview').load("preview", {text:$("#email_blast_message").val()});  

+

  });

+

:textile

+

  #{@text}

+

  def translate

+

    @text = params[:text]

+

    render "shared/translate"

+

  end

+

+

Which is a super fast way of ajaxing the text-area text to an ActionController::Metal action, which faster than thought possible, returns the HTML output version of the Textile markup’d text! Its incredibly fast since I’m only using Metal rather than a full ActionController object. Plus its very simple in what it does, so I’m pretty happy with how it worked out. Check out what it looks like!

+

[[posterous-content:yahilmbvnDwdtAofxsqb]]

diff --git a/_posts/2010-09-01-brand-new.html b/_posts/2010-09-01-brand-new.html new file mode 100644 index 000000000..e28a4dd36 --- /dev/null +++ b/_posts/2010-09-01-brand-new.html @@ -0,0 +1,18 @@ +--- +layout: post +title: Brand new +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047073980/brand-new +posterous_slug: post/1047073980/brand-new +--- +

First job as a freelance web designed is now out in the wild.

+ +

LowCountry Community Church

+ +

[[posterous-content:uucEeeyutwfyFFshgEza]]

+ +

Still wrestling between domain name providers to transfer the old domain over (from NetworkSolutions to GoDaddy). Just as a technical FYI, NetworkSolutions currently does not allow setting of a single A Record name (say www) to multiple ip addresses (as is needed with Heroku via this article.

+ +

Still work to be done in allowing the end admins to be able to customize the copy without my intervention, but that’ll come. This weekend was the deadline to make the switch over from their old static page site to mine. Not too bad….

diff --git a/_posts/2010-09-01-collaboration.html b/_posts/2010-09-01-collaboration.html new file mode 100644 index 000000000..3d8101d33 --- /dev/null +++ b/_posts/2010-09-01-collaboration.html @@ -0,0 +1,26 @@ +--- +layout: post +title: Collaboration +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047921500/collaboration +posterous_slug: post/1047921500/collaboration +--- +

[[posterous-content:oAnGiCmbEurGFrJuoIBi]] 

+

&

+

[[posterous-content:FCnppHpryfdqiJtvltbr]]

+ +

I’m very pleased to announce a new collaboration project between HungerAndthirst Design and TGStudios:

+ +

Peerdrum.

+ +

It’s a really slick app that works for companies looking to move more of their employees remote and for employees tired of working from a normal office.

+ +

Trust is usually a big issue for employers when evaluating whether or not to allow employees to work remotely. People work remote all the time, but for many smaller businesses and companies that rely on untested, untrusted contractors, it’s hard to ensure that people aren’t charging you for hours they really didn’t work or that employees aren’t getting paid for hours spent doing non-work related activities.

+ +

The idea behind Peerdrum is that employers should be able to rest easy that their employees are working while on the clock, and employees won’t have to worry about tracking time manually. The app will monitor the employees activities and send screenshots and hours logged back to the project supervisors. Want to jump off the clock, browse Facebook for a while? No problem, just pause the screenshot app on your desktop, spend some you-time, then start it back up again when you’re ready to start working again. Nothing gets logged when you’re off the clock, so all your private data is still untouched and your supervisors won’t have to worry about you playing around on their dime!

+ +

[[posterous-content:jgbpjnvjmwJxJFzoxuqc]]

+ +

The initial info can be found at the evangelism site for the project, Peerdrum.com. The project it set to launch sometime in August and I’m really excited to show this baby off. It’s some stuff I’ve never done before and its a pretty slick idea on the whole. Stay tuned…

diff --git a/_posts/2010-09-01-cool-ruby-plugin-for-the-day.html b/_posts/2010-09-01-cool-ruby-plugin-for-the-day.html new file mode 100644 index 000000000..511e1b4f2 --- /dev/null +++ b/_posts/2010-09-01-cool-ruby-plugin-for-the-day.html @@ -0,0 +1,16 @@ +--- +layout: post +title: Cool ruby plugin for the day... +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047058644/cool-ruby-plugin-for-the-day +posterous_slug: post/1047058644/cool-ruby-plugin-for-the-day +--- +

Ruby Progressbar: This little bugger is a really simple way to create a textual progress bar for your long-running ruby scripts. I’m working on a script that takes about 30-45 minutes to run. It does some file parsing and then querying of a web service based on the parsed values for reconciliation purposes. The thing is, the web service query takes about 3-5 seconds per query and we’re querying over 7000 records (which means we’d be talking about 7.5 hours to do this if we weren’t also doing some multithreaded magic).

+

I’m actually planning to run this thing in a few different ways out of personal curiousity. I want to run this in Ruby 1.9.1 with threads and compare that with the time it takes to fork individual processes out for each query. Right now, I’m running it with 15 threads (although I haven’t worked to figure out what an optimum number of threads is).

+

I also want to see what happens if I load up MacRuby’s implementation of Ruby 1.9.0 and run it under it’s GCD support to see how that compares against the other two. Although I suspect if I’m forking out individual processes, the GCD would be handling those processes anyways? I’m not intimate with the workings of GCD normally or under MacRuby. Time to research!

+

But for now, a screenshot of the progress bar thinger for Ruby scripts. Implementation details are on the script’s website.

+

[[posterous-content:FetjClCyChtFgJIhGDke]]

+

Update:

+

Totally forgetting what I learned in parallels class with the idea of trying to use Ruby’s Process library. Process doesn’t have a clean way to aggregate results from processes. It simply spawns off another Ruby instance, runs the given block and exits. So it looks as though if I want to create a concurrent execution version, I will need to either use a whole separate process spawning library… or MacRuby’s GCD. I think we’ll give GCD a chance first.

diff --git a/_posts/2010-09-01-css-nav.html b/_posts/2010-09-01-css-nav.html new file mode 100644 index 000000000..f7c3bc3af --- /dev/null +++ b/_posts/2010-09-01-css-nav.html @@ -0,0 +1,28 @@ +--- +layout: post +title: CSS Nav +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047927728/css-nav +posterous_slug: post/1047927728/css-nav +--- +

Okay, so maybe this is news to just me, but I’ve often struggled in the past to get a good looking login/admin control nav that sits neatly up in the top, right-hand corner of the screen without resorting to the unsightly, hard-to-control css float property. Float tends to bounce, jump, wrap, move at bad times… like during client demos. Ehem.

+ +

[[posterous-content:DDsbprrtgruzBCuyJHig]]

+ +

So I found a rather elegant solution from a few other sites. Really all it is is setting two divs:

+ +

+

<div style="position:relative" ...>

+

    <div style="position:absolute; top:0; right:0;"

+

        Nav Item 1

+

        Nav Item 2

+

    <div>

+

<div>

+

+

The first div simply ensures that the children div respect the relative positioning that you have it in the page. For example, in the image above, the nav is actually sitting down about 200 pixels and left about 400 from the absolute upper right corner. But the absolute tag isn’t pushing it to the top right because the relative div around the nav tag keeps it in place. This works for any browser I’ve tried it in and is not subject to strange float rules from browser to browser or getting displaced by other block elements.

+ +

Enjoy!

+ +

Anyone else have other methods out there for doing this?

diff --git a/_posts/2010-09-01-details.html b/_posts/2010-09-01-details.html new file mode 100644 index 000000000..dac977ff9 --- /dev/null +++ b/_posts/2010-09-01-details.html @@ -0,0 +1,21 @@ +--- +layout: post +title: Details +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047856421/details +posterous_slug: post/1047856421/details +--- +

And the Hastie saga continues… now have Facebook auth and event posting/updating working. Also got the Twitter auth working but haven’t had the time to get the event updating done yet. Should be really simple. Some cool details:

+ +

+[[posterous-content:meGmgaDjdszrujftsvIi]]

+

+And a small little friendly feature to let you know, Hey, you’re socially aware! Good for you! +

+

+[[posterous-content:xrcotgpnIbiecHprCzqf]]

+

+Small progresses. Always rewarding. Check it out for yourself at Hastie-app.com +

diff --git a/_posts/2010-09-01-development-to-the-max.html b/_posts/2010-09-01-development-to-the-max.html new file mode 100644 index 000000000..e313a0718 --- /dev/null +++ b/_posts/2010-09-01-development-to-the-max.html @@ -0,0 +1,15 @@ +--- +layout: post +title: Development to the max... +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047783983/development-to-the-max +posterous_slug: post/1047783983/development-to-the-max +--- +

Well, not only do I play the full-time 9-5er, but I launch new projects in the off time. I have so many side projects its not even that funny anymore. I need to settle down and work on one. The problem is that some of them involve a lot more work and time learning than I have time to do. Sometimes I just want a project that I can get into and go with rather than having to learn as I go. Ruby/Rails is my primary mode of development, so I’ve been trying for some time to think up a good project.

+ +

Tonight it finally happened. So now will you please welcome Hastie! The new way to schedule things. Actually its not new, but its got a lot of potential. I’m always a big fan of single source applications that tie a bunch of resources into a single go-to location. Now I get to try to write one. Lots of ideas in the works for this, most of which will take sometime, or well I suppose they might not since there is already so much functionality out there for some of the features I want to add in. Like Facebook and Twitter for example already have good and diverse APIs that people have written Ruby wrappers for. So I dunno how long it might take to add all the stuff I have planned for it, but its a project. A good one. Check it out, browse around, tell me what you think and if you recognize the design schema from any other projects!

+ +

Checkout Hastie!

+

[[posterous-content:flqGdjtlpEyydxpEfAEw]]

diff --git a/_posts/2010-09-01-dynamic-form-generation.html b/_posts/2010-09-01-dynamic-form-generation.html new file mode 100644 index 000000000..5df130481 --- /dev/null +++ b/_posts/2010-09-01-dynamic-form-generation.html @@ -0,0 +1,103 @@ +--- +layout: post +title: Dynamic form generation +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047893587/dynamic-form-generation +posterous_slug: post/1047893587/dynamic-form-generation +--- +

Interesting app for Hargray Communication this past few weeks. I was asked to design a dynamic survey application that is built, more or less, based on a configuration stored in a database. This is so we can have a front-end application in Oracle that allows, say, non-technical marketing people to build a survey from a form, and then it generates a set of web pages allowing the customer to take the survey.

+ +

So that’s what I did:

+

[[posterous-content:DAxJvyyvAIrBsboDdAzr]]

+ +

This is a fairly simple app on the whole. Solely purposed to allow customers to take surveys based on a link we send them. So why am I mentioning this?

+ +

Because dynamically generated apps are cool!

+ +

I originally did this with helpers that took the name of the question type (something like “single_choice” for radio buttons, “comment” for a multiline text area, etc.) and then an eval statement to call the right helper method. That made for some messy code in the helper. So I switched to using a set of partials instead, with the same names as what I have in the example above. This way I can do:

+ +

+

+

+

+

= render "shared/#{question.survey_question_type["sqy_code"].downcase}", :locals => {:form => f, :quest => question}

+

+

and get show the question. The radio-button partial looks simply like:

+ +

+

- locals[:quest].survey_answers.each do |a|

+

  = locals[:form].radio_button(locals[:quest]["sqt_name"].gsub(/(-|\s)/, "_").downcase.to_sym, a["saw_uid_pk"])

+

  = a["saw_web_answer"]

+

+

Look a bit weird? I’m also doing some crazy stuff for the answer generated from this form. Since I am in oracle and the data is stored using PLSQL procedures (not my choice), I can’t use the standard ARObject.save method. In fact, I’m not using ActiveRecord as the class base at all. Its actually using ActiveModel and acting like its an ActiveRecord object with a few fake methods to allow me to use it in the form_for helper and get standard AR validations. The class looks like:

+ +

+

+

class AnswerValidator < ActiveModel::EachValidator

+

  def validate_each(record, attribute, value)

+

    record.errors[attribute] << (options[:message] || "question not answered") if value.nil? and !record.errors[attribute].include? (options[:message] || "question not answered")

+

  end

+ +

end

+ +

class Answer

+

  include ActiveModel::Validations

+

  include ActiveModel::AttributeMethods

+ +

  attr_accessor :id

+ +

  def to_param

+

    (id = self.id) ? id.to_s : nil

+

  end

+ +

#   def to_key

+

#     persisted? ? nil : [id]

+

#   end  

+ +

  #force to always be new

+

  def new_record?

+

    true

+

  end

+ +

  def set_attributes(values)

+

    values.each do |k,v|

+

      self.send "#{k}=".to_sym, v

+

    end

+

  end

+ + +

  def initialize(questions=nil)

+

    # create accessor methods

+

    if questions

+

      questions.each do |q|        

+

        name = q["sqt_name"].gsub(/(-|\s)/, "_").downcase.to_sym

+ +

        class_eval {

+

          attr_accessor name

+

          validates name, :answer => true if q['sqt_required_fl']

+

        }

+

      end    

+

    end

+

  end  

+ +

end

+

+

As you can see, I have an AnswerValidator class that does the evaluation of the answers returned to make sure all the required which is set to validate for every record where q[‘sqt_required_fl’] is true. (This is the db column that flags if the question requires an answer to be complete).

+ +

Now the class_eval section actually dynamically creates a set of attr_accessor methods from the database fields so that I can access the fields as methods (necessary to make the validations form_for driven, since form_for looks for a method (object.attr_name) instead of a hash value (object[:attr_name]). Note: there may be a better way to do this?. Post in the comments if you know how.

+ +

The new_record? and to_param methods are needed by form_for as well to figure reference the object and to figure out if it is updating an existing record or creating a new one. The commented code to_key is what to_param becomes in Rails 3.0.0.beta3 (using beta currently).

+ +

Because I found issues…

+ +

…with this app when pushing new code in dev and not restarting the app. Why? Because of what is noted in this thread. Because the forms are being dynamically generated, they are relying on resources created at runtime. If I push new code to my dev server while the app is still running, the classes are dynamically regenerated. Which in turn means all the resources the dynamic form was relying on are now gone and weird call tree errors start showing up. So the real solution to this is in the config/environments/development.rb file, simply change:

+ +

+

+

config.cache_classes = false

+

#to

+

config.cache_classes = true

+

+

Just don’t forget to restart the app every time you push new code :)

diff --git a/_posts/2010-09-01-ie-gotcha-3.html b/_posts/2010-09-01-ie-gotcha-3.html new file mode 100644 index 000000000..327d8aece --- /dev/null +++ b/_posts/2010-09-01-ie-gotcha-3.html @@ -0,0 +1,20 @@ +--- +layout: post +title: ! 'IE Gotcha #3' +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047070622/ie-gotcha-3 +posterous_slug: post/1047070622/ie-gotcha-3 +--- +

Continuing the rant…

+ +

So I already knew IE7 didn’t respect the inline-block property. That sucks. Its really hard to do some types of layout without that property. So what to do?

+ +

Believe it or not, this hack works (from flipc.blogspot.com). Just add:

+ +

zoom: 1; *display: inline; _height: 30px;

+ +

to the end of your styling for the element you need inline-block to work for. It doesn’t screw with Firefox or even IE8 (which handles inline-block just fine…).

+ +

So moral of the story is still don’t use IE. Period. Solution one is Mozilla’s Firefox. Solution 2 is at least ie6update.com

diff --git a/_posts/2010-09-01-mallassi-project-story.html b/_posts/2010-09-01-mallassi-project-story.html new file mode 100644 index 000000000..df873cf6f --- /dev/null +++ b/_posts/2010-09-01-mallassi-project-story.html @@ -0,0 +1,10 @@ +--- +layout: post +title: ! 'InfoQ: The Story of a Project' +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/articles/mallassi-project-story +posterous_slug: articles/mallassi-project-story +--- +http://www.infoq.com/articles/mallassi-project-story?utm_source=twitterfeed&a...

Interesting take on how project lifecycle has changed

diff --git a/_posts/2010-09-01-moving-and-archiving.html b/_posts/2010-09-01-moving-and-archiving.html new file mode 100644 index 000000000..b9d8443cd --- /dev/null +++ b/_posts/2010-09-01-moving-and-archiving.html @@ -0,0 +1,16 @@ +--- +layout: post +title: Moving and archiving. +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047033944/moving-and-archiving +posterous_slug: post/1047033944/moving-and-archiving +--- +

I’ve finally decided to break down and not maintain my own blogging engine. Oddly enough, almost all Rails tutorials are all about “how to create a blog in under 15 minutes” but the maintaining one is actually not worth it when there are so many worthy blogging webapps out there. This is a cool one. Its like the awesome, non-bloated Wordpress cousin.

+ +

…INITIATE IMPORT…

+ +

hungerandthirstdesign.com/blogs will now become blog.hungerandthirstdesign.com

+ +

Thanks for watching!

diff --git a/_posts/2010-09-01-my-take.html b/_posts/2010-09-01-my-take.html new file mode 100644 index 000000000..9a52b0534 --- /dev/null +++ b/_posts/2010-09-01-my-take.html @@ -0,0 +1,31 @@ +--- +layout: post +title: My take +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047833975/my-take +posterous_slug: post/1047833975/my-take +--- +

First up: Apple and the 4g phone “leak”

+

+I think it was deliberate, although part of me is pretty sure it wasn’t. Not because Apple needs the press, but partly because it comes in a fairly vicious time for Apple and developers after the 3.3.1 clause in the SDK EULA. Maybe something to keep the general public happy and interested. So for what it’s worth, Apple, keep your stuff under control. Those kinda shenanigans melt servers with blog posts, feeds, and basically clog the interwebs. Just.Don’t.Do.It. +

+

+Better yet, what if this is another Gizmodo hoax? Faux-iPhone from China maybe? Some kickass design work from the graphics design department? +

+

Next: Gizmodo

+

+Companies or services (or whatever you call a group like the one behind Gizmodo) run on a business model that revolves solely around numbers. They’re really in the business of creating content around their ads. Ads is really the major way a blogging or news service makes its revenue, especially on the web with a non-subscriber based model. So it makes sense they would go after the biggest news story possible. I can’t blame them. Sure they probably could have done it better, sure they probably screwed Powell over. +

+

+The way I see it though, any company driven by “followers” rather than the quality of the services they can provide is already losing. What’s the point? Tabloidism is a dumb way to make a living. I’m not even sure I’d call what Gizmodo does a “service”. Its putting out content. Really low quality (and probably unedited) content… What’s more is when they find a story like this they talk it to death. Somebody kill it already. +

+The point is, if we lose site of service quality for numbers, we end up like McDonalds: Everywhere serving zillions but its not a place you’re proud to work. Its not something you’re proud to say you participated in. +

+I do have to say, the hardliner Apple geeks love ripping on Gizzie and its really funny to see. Settle down yo. +

+

Lastly…

+

+Checkout the new site design! Shoot any comments to me at @plukevdh on Twitter. +

diff --git a/_posts/2010-09-01-new-designs.html b/_posts/2010-09-01-new-designs.html new file mode 100644 index 000000000..41edc954a --- /dev/null +++ b/_posts/2010-09-01-new-designs.html @@ -0,0 +1,22 @@ +--- +layout: post +title: New designs +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047054175/new-designs +posterous_slug: post/1047054175/new-designs +--- +

Working on some new site designs, some for the company I work for on a full time basis, others for my church or personal interests. The portfolio is coming along.

+

[[posterous-content:rbteptBEmfGgzcvtrpHB]]

+

The first one, called very simply: App Monitor, is a very minimal design for a small, yet very useful site I need at work. I wanted a quick, dirty, yet elegant way to view website status, to check if the site was up, and similarly with the actual server machine. So threw together some quick graphics and a light backend to support it (with some slick JS tricks) and here we go. Its not accessible to public use, since it is an internal app, but I was really happy with the design.

+ + +

Uploaded with plasq’s Skitch!

+

This site has a lot of really interesting ideas working on the back end to do all the site checks and is using some new OSS projects I haven’t had time to play with otherwise. Also of note, this is my first project using Rails 3. I already want to switch all my other projects over, although I’m not sure how practical that is. We shall see. I’ll try to blog about that experience more in the near future.

+ +

My other site project is a more professional look, without a lot of flashy elements, but a very clean, information driven design. This one you can access in the public domain, although it is far from complete.

+ +

LowCountry Community Church demo site

+ +

It is still very much a work in progress but I hope to have this site up and functioning as the primary site within the next few weeks, and I am taking over the site for my church. Should be an interesting project for sure.

diff --git a/_posts/2010-09-01-passenger-3-coming.html b/_posts/2010-09-01-passenger-3-coming.html new file mode 100644 index 000000000..e38330a62 --- /dev/null +++ b/_posts/2010-09-01-passenger-3-coming.html @@ -0,0 +1,25 @@ +--- +layout: post +title: Passenger 3 coming... +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047899789/passenger-3-coming +posterous_slug: post/1047899789/passenger-3-coming +--- +

From Phusion’s website:

+ +

»

+ +

The Road to Passenger 3: Technology Preview 1 – Performance

+ +

By Hongli Lai June 10th, 2010

+

It has already been two years since we’ve first released Phusion Passenger. Time sure flies and we’ve come a long way since then. We were the first to implement a working Ruby web app deployment solution that integrates seamlessly in the web server, and all the features that we’ve developed over time – smart spawning and memory reduction, upload buffering, Nginx support, etc – have served us for a long time. Nevertheless, it is time to say goodbye to the old Phusion Passenger 2.2 codebase….

+ +

(continue reading here)

+ +

Some stats from the post:

+ +

[[posterous-content:lnFHqquyBomczaDmFwCh]]

+ +

Its a fine day to be a rubyist/rails dev.

diff --git a/_posts/2010-09-01-peerdrum-live.html b/_posts/2010-09-01-peerdrum-live.html new file mode 100644 index 000000000..b90215fef --- /dev/null +++ b/_posts/2010-09-01-peerdrum-live.html @@ -0,0 +1,14 @@ +--- +layout: post +title: ! 'Peerdrum: Live!' +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047924533/peerdrum-live +posterous_slug: post/1047924533/peerdrum-live +--- +

Check it out! First major project to hit the big time. More press material coming soon!

+

peerdrum.com

+ +

[[posterous-content:bqrzlppghsfbnJaetket]]

+

[[posterous-content:eqpuormiicHbItHGnqdq]]

diff --git a/_posts/2010-09-01-podcasting-skip-filter-and-ha.html b/_posts/2010-09-01-podcasting-skip-filter-and-ha.html new file mode 100644 index 000000000..4ea0c84a8 --- /dev/null +++ b/_posts/2010-09-01-podcasting-skip-filter-and-ha.html @@ -0,0 +1,64 @@ +--- +layout: post +title: Podcasting, skip_filter and Hastie progress +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047807839/podcasting-skip-filter-and-hastie-progress +posterous_slug: post/1047807839/podcasting-skip-filter-and-hastie-progress +--- +

Hastie update

+ +

Well an interesting weekend for sure. Hastie is still coming along. Did some really stupid stuff while working on it late one night whilst trying to integrate facebook connectivity. mmangino’s amazing Facebooker plugin for Rails works really well if you follow the instructions. Warning: DO NOT ADD “require ‘facebooker’” anywhere in the code. It creates traceless stack crashes which are never fun to debug at 2 in the morning. I ended up resetting my database in an attempt to solve it and deleted some important static data I was holding there. Still kicking myself over that one. In any case, Hastie is coming right along. I should be posting a major update later this week.

+ +

[[posterous-content:qvkknybAgwjmaIoDcICG]]

+ +

skip_filter

+ +

But this isn’t a post about Hastie. Wanted to post just some interesting things I learn about Rails as I go along on some various projects. Hargray Communication has a massive Oracle database that I have to write my apps against. Oracle is not particularly “Rails-friendly” although its more the fact that our db schema/naming conventions are not Rails approved. So it always takes a lot of wrangling. Why is this a problem? Well, because our user tables are set in stone, most of the nice authentication plugins out there (authlogic, restful-authentication) don’t work (at least by default) with our db schemas. Therefore, I end up writing a lot of custom authenticators. I suppose the cool thing to do would be to fork and mod something like authlogic to work with our databases, which thinking about that now, wouldn’t be too hard since it would mostly involve removing functionality. But anyways, I wrote a very simple auth system using almost exclusively filters. Take a look!

+

+

+

#application_controller.rb

+

  before_filter :deny_access, :unless => :logged_in?

+

  private

+

  def deny_access

+

    flash[:notice] = "You cannot access that page without logging in first."

+

    redirect_to "/"

+

  end

+

  def logged_in?

+

    session[:username]

+

  end

+

#session_controller.rb

+

  skip_filter :deny_access

+

+

The session controller can simply skip the auth filter on all of its actions while ensuring that every other action in the application checks that we have a user logged in. We could add conditions to the skip filter if there were other session actions that only an authorized user would be allowed to access, but all I do in the session controller for now is login and logout. I previously hadn’t realized that skip_filter worked on methods outside of its own controller. So that’s pretty slick.

+ +

Podcasting

+ +

Added podcast hosting/upload ability to theLowCountry Community Church website yesterday. Took a few hours, but it really wasn’t super difficult.

+

[[posterous-content:bDBGxbjkylFkfzCbfEap]]

+

The image above is a placeholder for what a podcast entry on the page looks like. You can actually subscribe to the feed in iTunes under the “Advance > Subscribe to Podcast” option. I expect it will take some time to show up in the iTunes store podcast listing (submitted yesterday). I didn’t realize until I started researching this project, but a podcast feed is just an XML rss feed. Ours is found here. It just involves a mess of description tags. Our files are actually being hosted on Amazon S3 (for less than 5 cents a month) and I’m pretty stoked with how easy the whole thing was. A lot of the more dynamic content is stored there simply because its so easy to upload the images/audio there. The aws-s3 gem is fantastic for accomplishing the uploading/linking files. A bit of a gotcha in the whole process is ensuring that the files are accessible to the general public with no authentication required. otherwise iTunes complains. Its pretty simple though. My upload and link generating methods look like this:

+ +

+

  def get_podcast_image(name)

+

    # :authenticated => false ensures that the urls

+

    # returned for the object don't have the weird auth keys included

+

    AWS::S3::S3Object.url_for(name, "podcast_images", :authenticated => false)

+

  end

+ +

  def self.upload_podcast_image(name, url)

+ +

    if !url.nil? and !name.empty?

+

      # in order for the :authenticated => false to work, 

+

      # we need to upload the files with access opened to the public. 

+

      # :access => :public_read accomplishes this

+

      AWS::S3::S3Object.store(name, open(url), "podcast_images", :access => :public_read)

+

      true

+

    else

+

      "Missing information with this podcast."

+

    end

+

  end

+

+

In any case, this works beautifully and hopefully before much longer, you’ll be able to get all the sermons from LowCountry Community Church online or through iTunes. And semi-awesome side note, the audio player seen in the screenshot above is the html5 “audio” tag at work! Works really well in Safari and Chrome. So far no Firefox love. Hopefully that’ll come soon.

+ +

I realize I have no comments section here yet. I’ll work at that as well as an RSS feed soon, but if you have any comments, please email me at hungerandthirst[at]gmail[dot]com.

diff --git a/_posts/2010-09-01-rails-3-is-coming.html b/_posts/2010-09-01-rails-3-is-coming.html new file mode 100644 index 000000000..165f307d3 --- /dev/null +++ b/_posts/2010-09-01-rails-3-is-coming.html @@ -0,0 +1,13 @@ +--- +layout: post +title: Rails 3 is coming... +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047895838/rails-3-is-coming +posterous_slug: post/1047895838/rails-3-is-coming +--- +

Beta 4 is out!

+

Also a big step for Bundler which just hit 1.0.0.beta1

+

So much coming soon!

+

[[posterous-content:wraaolohjjjvFgiCkffg]]

diff --git a/_posts/2010-09-01-rails-3-progression.html b/_posts/2010-09-01-rails-3-progression.html new file mode 100644 index 000000000..d2163234e --- /dev/null +++ b/_posts/2010-09-01-rails-3-progression.html @@ -0,0 +1,14 @@ +--- +layout: post +title: Rails 3 progression +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047827560/rails-3-progression +posterous_slug: post/1047827560/rails-3-progression +--- +

[[posterous-content:DhJBjbofvmtBtakfndrr]]

+ +

Check it out!! Converted this site to Rails 3 beta 3. Very cool. Way to go Rails team. Keep up the work.

+ +

UPDATE: Apparently Heroku does not actually allow beta 3 yet. I think it require Ruby 1.9.2 which is still in pre-release state. So close…

diff --git a/_posts/2010-09-01-servers.html b/_posts/2010-09-01-servers.html new file mode 100644 index 000000000..19d272eb1 --- /dev/null +++ b/_posts/2010-09-01-servers.html @@ -0,0 +1,14 @@ +--- +layout: post +title: Servers +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047047187/servers +posterous_slug: post/1047047187/servers +--- +

Update on our hosting situation. I’m very pleased to announce the final hosting location for our site here and for future projects will be Heroku.com. I had used them once before in the past before I had a set of domain names and now having that along with an updated set of features from Heroku’s dev team, I’m very happy to be hosting my stuff here.

+ +

Before the move, I had been trying to host on Bluehost. They provide an excellent spread of services and options for a great cost, however, they aren’t very Rails specific, so trying to get setup there made for a fairly volatile combination of config files and weird apache hacks. So switching to Heroku should keep the site clean and stable.

+ +

Because of Heroku’s very progressive support of Rails, I’m actually hoping to get some experimental stuff up here to demo Rails 3. This site may even move to R3 before much longer. Just glad to finally have this site, namely the blog, up and running. Stay tuned for current projects and other updates!

diff --git a/_posts/2010-09-01-steam-for-mac.html b/_posts/2010-09-01-steam-for-mac.html new file mode 100644 index 000000000..42c350dfe --- /dev/null +++ b/_posts/2010-09-01-steam-for-mac.html @@ -0,0 +1,20 @@ +--- +layout: post +title: Steam for Mac +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047851328/steam-for-mac +posterous_slug: post/1047851328/steam-for-mac +--- +

I’ve been a Steam user for a while now, but as a Mac user, its always been fairly frustrating to make games happen on my Mac. Always an issue. But now, today, finally… +

+

+Its here. +

+[[posterous-content:zBhbDEEIEGCzjbGDzHqH]]

+First impressions are just that the interface is really well done. Very clean, intuitive, and it works! Doesn’t look like the store is up on the client yet, but it exists on Steam’s site. Mostly some slick indie games and Portals!! Now we just wait for Portals 2… +

+

+Also looks like they’ll be releasing new games for the Mac every Wednesday +

diff --git a/_posts/2010-09-01-they-said-it-couldnt-be-done.html b/_posts/2010-09-01-they-said-it-couldnt-be-done.html new file mode 100644 index 000000000..0d6c971db --- /dev/null +++ b/_posts/2010-09-01-they-said-it-couldnt-be-done.html @@ -0,0 +1,16 @@ +--- +layout: post +title: They said it couldn't be done... +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047044044/they-said-it-couldnt-be-done +posterous_slug: post/1047044044/they-said-it-couldnt-be-done +--- +

Actually nobody said that. Or maybe I did. In any case…

+ +

After much delay and a bit of big talk about trying to launch this deal for real, I at least have the blog together. I’m still very much in the process of putting together the site design on the whole, but I’m liking where it is. Authentication bits are now working although being logged in doesn’t really do anything for you… so yeah. Not much point to that yet. Hang in there though, I promise it will be worth it. 

+ +

Also looking to soon add ways for you to contact me here at HungerAndThirstDesign.com, as currently, our only method of contact is really Twitter (HaTD). So please drop us a line and we’ll get back to you that way, push updates, or generally tell you when you’ll be able to contact us for reals.

+ +

Lastly, just want to say I’m very excited about the upcoming release of Rails 3, and hoping that the Rails team can make good on the hints to push it out late February. Will definitely be looking to make this and future sites work better, faster and more efficient on the new framework. Pumped!!

diff --git a/_posts/2010-09-01-this-will-only-make-things-wo.html b/_posts/2010-09-01-this-will-only-make-things-wo.html new file mode 100644 index 000000000..194c64832 --- /dev/null +++ b/_posts/2010-09-01-this-will-only-make-things-wo.html @@ -0,0 +1,21 @@ +--- +layout: post +title: This will only make things worse... +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047935800/this-will-only-make-things-worse +posterous_slug: post/1047935800/this-will-only-make-things-worse +--- +

[[posterous-content:roGtybwalHcGBDmDamkz]]

+

I already entered “:wq” in way too many other editors. This certainly isn’t going to help.

+ +

Starting up with vim as my primary editor. Going to take some time to get used to, but I really like it so far. I always wanted to use keyboard only navigation…

+ +

Some helps I’ve found along the way are:

+ +

Vim

+

Peepcode screencasts (I and II)

+

Ubuntu fix for Ack

+

Yehuda Katz’s thoughts.

+

This will probably be a fluid list, so stay tuned!

diff --git a/_posts/2010-09-01-typekit.html b/_posts/2010-09-01-typekit.html new file mode 100644 index 000000000..02fba8c9e --- /dev/null +++ b/_posts/2010-09-01-typekit.html @@ -0,0 +1,12 @@ +--- +layout: post +title: Typekit +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047842280/typekit +posterous_slug: post/1047842280/typekit +--- +

As some may have noticed from the logo in the bottom of this screen, I use Typekit for some font design for my web projects. They are a very cool service adding a lot of “cool-factor” to any web project. Yet another reason to love being a web developer.

+ +

And as-per this latest blog post, they just opened up their Portfolio account to use unlimited fonts for unlimited sites. Pretty cool. I had often wondered if I would be able to use them beyond a few sites as I get into more and more freelance work. Very good to know they’ve been concerned with this too. Thanks Typekit! Keep up the amazing work.

diff --git a/_posts/2010-09-01-vim-thoughts.html b/_posts/2010-09-01-vim-thoughts.html new file mode 100644 index 000000000..be3c1be58 --- /dev/null +++ b/_posts/2010-09-01-vim-thoughts.html @@ -0,0 +1,21 @@ +--- +layout: post +title: Vim Thoughts +published: true +date: 2010-09-01 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1047946228/vim-thoughts +posterous_slug: post/1047946228/vim-thoughts +--- +

So I mentioned last week that I was making the switch to VIM. Well here we are, a week later, and where are am I so far? Much better than expected. I installed fresh VIM on my linux dev machine and, since I haven’t found a dependable editor on linux that wasn’t slow as hell or almost totally useless for my customary workflow, it was fairly easy not to be tempted to revert to my old editor, since I didn’t really have one on Linux. Mac still pulls me in the way of TextMate or Coda (which is still one of my favorites of all time), but even there, for quick edits, and even hardcore refactoring, its getting really hard to be without VIM’s super-extensive and extendable keyboard shortcuts.

+ +

It’s definitely a learning curve and I’m still very much learning and trying to remember what does what and what will make this function faster, how I can bounce about, replace, update code better. It takes a lot of memory power, but starting out slow, you kinda just learn as you go along. :h gives me plenty of info about things I’ve seen people do or want to learn how to do. The built in help is probably one of the best things ever. Every editor should come with this much help before it can be released. I’m srsly you guys…

+ +

I’m generally a fan of one input device if I can help it, and this is definitely where VIM shines for me. Even in MS Word, I start typing :w to save instead of Ctrl-S… But now that I’m getting the swing of the navigation tools within VIM, I’m getting pretty versatile in moving around the code really quickly. A few plugins I would recommend using early on (regardless of what language you’re working with) to aid in the migration are:

+ +

Pathogen – Simple plugin installation just by dropping vim plugins into a single folder.

+

NERDTree – A TextMate-like drawer/file browser for VIM. Splits browsing files into a separate buffer for quick perusal of projects and the like.

+

BufExplorer – Lets you jump around your buffers very quickly. Not necessary, but I use this one to bounce around files a lot faster. Almost like tabs for your editor (without using VIM’s built-in tabs).

+

VCSCommand if you use SVN or a wide variety of other SVC systems. Git is included in this pkg, but I like to use vim-git and vim-fugitive instead).

+

Ack – Better than grep. All you need to know.

+

There are half a dozen others I use for Rails/Ruby development, but these are the mainstays. I am in VIM to stay for now. At least until TextMate2 comes out…

diff --git a/_posts/2010-09-02-amazing-how-much-turnaround-s.html b/_posts/2010-09-02-amazing-how-much-turnaround-s.html new file mode 100644 index 000000000..3fb39b459 --- /dev/null +++ b/_posts/2010-09-02-amazing-how-much-turnaround-s.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1050909598/amazing-how-much-turnaround-since-the-iphone-4 +posterous_slug: post/1050909598/amazing-how-much-turnaround-since-the-iphone-4 +--- +

amazing how much turnaround since the iphone 4 debut. people went crazy. now everyone is complaining…

diff --git a/_posts/2010-09-02-eating-large-chunks-of-banana.html b/_posts/2010-09-02-eating-large-chunks-of-banana.html new file mode 100644 index 000000000..e68248de3 --- /dev/null +++ b/_posts/2010-09-02-eating-large-chunks-of-banana.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1050911223/eating-large-chunks-of-banana-bread +posterous_slug: post/1050911223/eating-large-chunks-of-banana-bread +--- +

eating large chunks of banana bread

diff --git a/_posts/2010-09-02-i-have-so-much-data-to-export.html b/_posts/2010-09-02-i-have-so-much-data-to-export.html new file mode 100644 index 000000000..4b15dfbe8 --- /dev/null +++ b/_posts/2010-09-02-i-have-so-much-data-to-export.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1050911334/i-have-so-much-data-to-export-oh-wait +posterous_slug: post/1050911334/i-have-so-much-data-to-export-oh-wait +--- +

I have so much data to export!! oh wait… http://is.gd/ePnMc

diff --git a/_posts/2010-09-02-in-other-news-redis-is-actual.html b/_posts/2010-09-02-in-other-news-redis-is-actual.html new file mode 100644 index 000000000..c9532b5fb --- /dev/null +++ b/_posts/2010-09-02-in-other-news-redis-is-actual.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1054381647/in-other-news-redis-is-actually-pretty-awesome +posterous_slug: post/1054381647/in-other-news-redis-is-actually-pretty-awesome +--- +

in other news: Redis _is_ actually pretty awesome.

diff --git a/_posts/2010-09-02-note-to-self-never-allow-upda.html b/_posts/2010-09-02-note-to-self-never-allow-upda.html new file mode 100644 index 000000000..3d444a688 --- /dev/null +++ b/_posts/2010-09-02-note-to-self-never-allow-upda.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1050910739/note-to-self-never-allow-updates-from-verizon +posterous_slug: post/1050910739/note-to-self-never-allow-updates-from-verizon +--- +

note to self: never allow updates from Verizon directly. derooted my droid… :(

diff --git a/_posts/2010-09-02-rt-jasonmauer-ping-the-itunes.html b/_posts/2010-09-02-rt-jasonmauer-ping-the-itunes.html new file mode 100644 index 000000000..f22be3bd2 --- /dev/null +++ b/_posts/2010-09-02-rt-jasonmauer-ping-the-itunes.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1054381637/rt-jasonmauer-ping-the-itunes-social-network +posterous_slug: post/1054381637/rt-jasonmauer-ping-the-itunes-social-network +--- +

RT @jasonmauer: Ping, the iTunes social network. like I want to spend more time in iTunes. it’s like having a social network in prison.

diff --git a/_posts/2010-09-02-rt-wilshipley-itunes-we-wont-.html b/_posts/2010-09-02-rt-wilshipley-itunes-we-wont-.html new file mode 100644 index 000000000..cedeb7e26 --- /dev/null +++ b/_posts/2010-09-02-rt-wilshipley-itunes-we-wont-.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1054381652/rt-wilshipley-itunes-we-wont-be-happy-until-it +posterous_slug: post/1054381652/rt-wilshipley-itunes-we-wont-be-happy-until-it +--- +

RT @wilshipley: iTunes: We won’t be happy until it embeds Virtual Basic.

diff --git a/_posts/2010-09-02-well-thats-done.html b/_posts/2010-09-02-well-thats-done.html new file mode 100644 index 000000000..3b2698d91 --- /dev/null +++ b/_posts/2010-09-02-well-thats-done.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1050911419/well-thats-done +posterous_slug: post/1050911419/well-thats-done +--- +

well that’s done :)

diff --git a/_posts/2010-09-02-why-does-everybody-and-everyt.html b/_posts/2010-09-02-why-does-everybody-and-everyt.html new file mode 100644 index 000000000..f2e20a34f --- /dev/null +++ b/_posts/2010-09-02-why-does-everybody-and-everyt.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1054381653/why-does-everybody-and-everything-need-social +posterous_slug: post/1054381653/why-does-everybody-and-everything-need-social +--- +

why does everybody and everything need social networking?????

diff --git a/_posts/2010-09-02-wow-browsers-are-getting-real.html b/_posts/2010-09-02-wow-browsers-are-getting-real.html new file mode 100644 index 000000000..53764d306 --- /dev/null +++ b/_posts/2010-09-02-wow-browsers-are-getting-real.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1050911032/wow-browsers-are-getting-really-freaking-intense +posterous_slug: post/1050911032/wow-browsers-are-getting-really-freaking-intense +--- +

wow, browsers are getting really freaking intense - http://is.gd/ePrQE

diff --git a/_posts/2010-09-08-a-tale-of-two-asyncs.html b/_posts/2010-09-08-a-tale-of-two-asyncs.html new file mode 100644 index 000000000..74e187c7d --- /dev/null +++ b/_posts/2010-09-08-a-tale-of-two-asyncs.html @@ -0,0 +1,27 @@ +--- +layout: post +title: A tale of two asyncs... +published: true +date: 2010-09-08 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1086573384/a-tale-of-two-asyncs +posterous_slug: post/1086573384/a-tale-of-two-asyncs +--- +

My latest project has been a really interesting one for being thrown back to fundamental programming problems mixed in with some really fascinating tools. First, some history.

+

[Partly excerpted from the project README]

+ +

There are a number of projects I can think of where a job that takes longer than a user ought to be waiting for a result (due to server timeout length or out of simple app responsiveness courtesy to the user). At the same time I need the result of the operation to be returned to the user.

+

Here’s a real-world example. I work with a set of legacy Oracle databases that stores much of our business logic as PLSQL procedures. Yes, this is not “The Rails Way™” but it’s the only way for the company I work for right now. Many of the procedures that I run as part of several of the applications I support can take on average one minute or more with a standard deviation of almost 2 minutes (with a forced timeout of 5 minutes). That’s kinda a long time to sit and wait on a web app regardless of what web app we’re talking about.

+ +

[[posterous-content:olburAEvomACaqifsxgl]]

+ + +

We don’t really want users sitting waiting for up to 5 minutes (at which it forces failure) unable to do anything or (even worse) hitting refresh or the action again. Especially bad when this can mean the HTTP server is getting backed up as more and more people run these long running processes.

+

Moreover, the users need to get response from the completed job before moving on. Most job processors (DJ, Resque) are setup for running jobs that do not require the result to be returned to the web app (think mass-mailers, queue population, image resizing). They just run and the output goes to a database, an inbox or a file server.

+

Kthxbye is an attempt to solve this problem. It is based heavily off of Resque, so much so that one might wonder why not just hack Resque or write a plugin for it? Simply put, I’m very interested in distributed programming. I wanted to try my hand at this. Secondly though, I haven’t had much time to play with these new NoSQL dbs such as Redis and this is a good opportunity to take it for a spin. Lastly, I’m interested in async and evented programming. 

+

One thing a number of the distributed/delayed/backend processors do is run things in separate processes. Resque in particular does this (pointed out only because I’ve dug into the Resque code more than others). I’m not a huge fan of spawning out additional processes. In Ruby, there are really two main ways to do asynchronous processing on a single machine:

+
    +
  1. The Parent/Children Model = Spawning multiple processes
    2. +
  2. The Fiber Model = Programmer-Controlled Threads
    3. +

Ruby has threads but due to the infamous GIL, Ruby’s solution to forcing thread-safety, we’re really locked out of threading like we ought to be (Queue up fun graphs and a better explanation). This is unless you run JRuby which uses Java’s much more satisfactory threading API. Firing up an additional ruby instance is pretty costly as it means you are running multiple instances of the Ruby interpreter. And why do that when we have threads good threads, albeit manually-managed threads. Rails is about to go fiber’d in 3.1, so why not Kthxbye? So part of this awesome hack project is to implement a distributed, “async”, delayed job processor using Redis and Ruby Fibers and to learn how to control my threads manually, something all good programmers should be able to do anyways.

+

Please check out my progress via the Github changelog and jump in with any suggestions, feedback or patches! Keep in mind this is still very early stage and experimental, but it could be awesome.

diff --git a/_posts/2010-09-10-because-i-told-you-before.html b/_posts/2010-09-10-because-i-told-you-before.html new file mode 100644 index 000000000..a39fa7235 --- /dev/null +++ b/_posts/2010-09-10-because-i-told-you-before.html @@ -0,0 +1,16 @@ +--- +layout: post +title: Because I told you before... +published: true +date: 2010-09-10 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1095348551/because-i-told-you-before +posterous_slug: post/1095348551/because-i-told-you-before +--- +

No, you can’t do that.

+

[[posterous-content:xnmGwuhEHtfruyjyDcxa]]

+

So I’m gonna do a bit of backtracking on what I told you before. After a good couple of days of futzing about with fibers, EventMachine and Processes and have made some interesting discoveries about how to go about things when working out concurrency.

+

Fibers are awesome, but they’re a bit tricky when it comes to using at a higher level around other libraries. Why? Because there is no good way to preempt fibers when it hits a blocking process… such as a long DB call, which is the whole point. In order to get that level of preemption, I’d have to make all the supporting libraries fiber aware as well. That’s a lot of work for this project. It would involve, at least in this instance, modifying Raimonds Simanovskis PLSQL adapter and probably the oracle_enhanced gem as well and that’s way beyond the scope of this project. Probably worth doing, just not what I’m working on at the moment.

+

So that leaves EM and Processes. I worked with EM for a bit and discovered its a bit tricky to use within an infinite loop. It works, just not as well as I’d like and without the level of concurrency or controllability I would like. Deferrables are nice, but it’s hard to spawn off multiple defers for each of the blocking processes that might be run in parallel. 

+

More importantly, I’ve decided that running multicore-able is going to be better that multi-threadable (which would only allow us to run on a single core in Ruby due to the GIL). The only way to do that in Ruby is, yep, you guessed it: Processes. And reading this article on Unicorn makes me believe using processes rather than threads/fibers or evented programming is a really respectable option. Perhaps the only option for web/network programming. Threads just don’t perform nicely in Ruby. Fibers don’t offer enough options or true concurrency (rather cooperative scheduling). Evented Programming doesn’t work as well for our architecture. 

+

So on to processes. This architectural decision has held up development of Kthxbye now for a couple of days, so now onward to greatness. Stay tuned.

diff --git a/_posts/2010-09-23-kthxbye-1-0-0.html b/_posts/2010-09-23-kthxbye-1-0-0.html new file mode 100644 index 000000000..8466e6d42 --- /dev/null +++ b/_posts/2010-09-23-kthxbye-1-0-0.html @@ -0,0 +1,23 @@ +--- +layout: post +title: Kthxbye 1.0.0 +published: true +date: 2010-09-23 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1174317195/kthxbye-1-0-0 +posterous_slug: post/1174317195/kthxbye-1-0-0 +--- +

Hey all. quite pleased to introduce the “final” version of my latest project, Kthxbye. As I wrote about in the last few blog posts, this gem allows you farm out long running processes that you need a result from to a separate processing engine so that you don’t have to tie up your web-server with long running processes.

+

This doubtlessly will have many revisions in the future but it took me so long to finalize the initial API that I wanted to solidify something for an initial release. This library is fully tested for Ruby 1.8.7 and 1.9.2. 

+

[[posterous-content:dgHmfgBHgoFreGqIlEDz]]

+

It has a pretty slick web front-end too. Lots of useful information to display as well as a live polling feature which refreshes the data on the screen every 2 seconds.

+

[[posterous-content:AwddHqhowJzfxtbzvAsc]]

+

It also gives you good indication as to what the status of the job is, when it’s completed, whether or not it succeeded or failed. This is all available through the API, and therefore, to your app as well.

+

[[posterous-content:sGofbgqHpChsgyDaacJu]]

+

The interface is built on Sinatra and should be pretty easy to integrate into your Rails app stack (made especially easy in Rails 3).

+

So how about some comments on the project? This is my first time doing any kind of major unit testing. Definitely will be doing more of this in the future. Incredibly helpful to know when I broke something earlier, caught so many more exceptions and edge cases I might have otherwise missed. Obviously that’s the point of unit testing, but it was cool to see just how helpful it was long term.

+

Where to get this sugary goodness?

+

Code or Gem

+

Documentation is soon to follow. In fact, it exists, it just needs to be better. Undocumented code is almost as useless as no code… 

+ +

UPDATE: Docs now available via rdoc.info, courtesy of github service hooks!

diff --git a/_posts/2010-09-25-dont-stay-in-bed-unless-you-c.html b/_posts/2010-09-25-dont-stay-in-bed-unless-you-c.html new file mode 100644 index 000000000..1aaf4b8cd --- /dev/null +++ b/_posts/2010-09-25-dont-stay-in-bed-unless-you-c.html @@ -0,0 +1,11 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-09-25 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1183766719/dont-stay-in-bed-unless-you-can-make-money-in +posterous_slug: post/1183766719/dont-stay-in-bed-unless-you-can-make-money-in +--- +
Don’t stay in bed, unless you can make money in bed.
+

George Burns (goodnight all. its 4 in the morning)

diff --git a/_posts/2010-09-25-site.html b/_posts/2010-09-25-site.html new file mode 100644 index 000000000..0a2235d6e --- /dev/null +++ b/_posts/2010-09-25-site.html @@ -0,0 +1,22 @@ +--- +layout: post +title: Site +published: true +date: 2010-09-25 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1183754580/site +posterous_slug: post/1183754580/site +--- +

Finally made the first of many updates to my site. Color scheme change is next. Or maybe the reskinning of this blog?

+

[[posterous-content:joHjmiixqBsCjuFAywmu]]

+

In any case, I’m now with webbynode. I figured if anything, my site ought to be the first one to make a switch as I’m getting a little discontent with Heroku’s availability some days. Moreover, their charges for additional services are getting outrageous. 

+

It’s not incredibly difficult to setup. Definitely some changes over Heroku, who I must say still have one of the best deploy workflows out there. But what is incredibly valuable is I now have root access to the server, which means I can do just about anything I want to it. I can kill it in cold blood if I want. Big. Fat. “shutdown -t now”, But lets not get hasty…

+

The big benefit is I can now run Node.JS, Redis, MongoDB, custom ruby tasks, cron jobs… anything I want… for free. Heroku is charging an arm and a leg for that privilege. 

+

So first is migrating everything from heroku. Not hard. Webby is not as well documented, but its not too hard using the preconfigured Linux stacks. They even have Rails 3 (in fact I hear they were one of the first web hosts to support the beta). 

+

A few minor complaints thus far. You can’t include your database.yml in the lineup or it complains. Also when using mysql, you are forced to use the old mysql gem rather than the shiny new mysql2 gem which has a lot going on. The fix for this currently is just to update it by hand post-install.

+

Besides that and some trouble figuring out how to ssh into the thing (simply due to lack of documentation), hungerandthirstdesign.com is running like a champ. Moreover, I can now run some other images processors (setup to thumbnail images for Peerdrum.com) that I had running on an EC2 instance over here now, which is pretty freaking awesome since EC2 costs me $60/mo. to run with no real usage out of it. So that’s a big deal.

+

Because of my work supporting Hargray’s custom linux rails server stack, I know how to do custom scaling and failover for these machines and while having Heroku’s slider based scaling system is nice, doing it my own way is nicer. 

+

New experiment too: Trying on Nginx for the first time. Running passenger on top of that for now. Seems like a really clean way to do web server configuration. I like. Seems fast enough. Consumes very little memory which is good on my lower-end node.

+

So all in all, I like Webbynode so far. I’ll be interested to start integrating more stuff into it like Redis or some other db backend. I guess in a sentence to sum it up: root access is awesome.

+ +

[[posterous-content:CFejrabuxquhwCGliICa]]

diff --git a/_posts/2010-09-30-messing-with-the-tubes.html b/_posts/2010-09-30-messing-with-the-tubes.html new file mode 100644 index 000000000..e99976ff8 --- /dev/null +++ b/_posts/2010-09-30-messing-with-the-tubes.html @@ -0,0 +1,41 @@ +--- +layout: post +title: Messing with the Tubes +published: true +date: 2010-09-30 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1215411972/messing-with-the-tubes +posterous_slug: post/1215411972/messing-with-the-tubes +--- +

So Kthxbye takes a turn to the even more experimental. The last big step of this whole project is to take the data coming back from the workers and notify the user that their data is ready and/or doing something on the client side with that data. If that sounds hard, the answer is actually under 100 lines of code (if you don’t count the supporting libraries). Its an incredibly versatile solution for asynchronous, server-driven page content updating. Push notifications for the web if you will.

+

However, this trip gets pretty crazy… and doesn’t work quite right yet, at least not in… dun-du-dun… Internet Explorer. But that’s another problem to solve as we go along (http://stackoverflow.com/questions/3825271/socket-io-node-js-rails-3-ie).

+

So there is now a second optional module to add to Kthxbye when using it behind a Rails application. I’ve designed a slick little notification widget that will be updated in realtime when the job gets picked up to be worked, once it completes or fails. All you’ll need to do in the end is include a CSS file, a javascript file and a

+

<div id='kthxbye'></div>

+

tag to your html or other markup and you’ll get this (or well, the first of the four blocks).

+

 [[posterous-content:gjerkuFGnkteaEdonGEb]]

+

Then once you startup a worker, it’ll change the colors of the lights based on the status of the job.

+

How does this work? Its a three layered system. Kthxbye uses Redis’ PUBSUB protocol to publish a message to the “job.*” channel, where “*” will be success|failed|active. There is then a Node.js server sitting in-between the Rails app and Redis that subscribes to the “job.*” channel waiting for messages to come down. The message currently only contains a single number with the original job id that it was queued under, this way when the clients receive the message, they can filter for what message they want on the client-side rather than opening up potentially infinite number of channels for each job, which as it appears from the Redis docs would ultimately slow down publishing to the messaging queues as per this specification: 

+
+

publish is O(N+M) where N is the number of clients subscribed to the receiving channel, and M is the total number of subscribed patterns (by any client). Psubscribe is O(N) where N is the number of patterns the Psubscribing client is already subscribed to.

+

(http://code.google.com/p/redis/wiki/PublishSubscribe)

+
+

So we go with three main channels: job.active, job.failed, job.success. There may be more in the future and someone can please feel free to correct me if my logic is wrong there.

+

So why the Node.js server in between? Why not poll ourselves from the app itself and update when we get the messages? A few reasons:

+
    +
  1. WebSockets are awesome. Incredibly lightweight and fantastic for not chewing up resources while we wait for content.
    2. +
  2. Node.js is an evented server implementation which is much cleaner when waiting for responses in a deferrable fashion. It also ties in really nicely with Socket.IO and Redis already.
    3. +
  3. Node.js is a lot less memory intensive (especially during idle time) and allows us to hold our polling mechanism in one location decreasing the load on Redis and our clients at the same time. This decreases load on both Redis and the Rails server allowing a server made for the sole purpose of handling these types of requests in an amazingly efficient fashion… do just that.
    4. +
  4. I needed a good reason to play with Node.js too. This was it.
    5. +

So why Socket.IO?

+
    +
  • It allows for clean failover in the event that one protocol is not supported by a browser. It supports WebSockets, Adoble Flash Sockets, ActiveX HTMLFile and xhr multipart and xhr polling requests. It successively falls through each type until one works. Obviously WebSockets is the best of these options, but its really handy to have the other protocols while other browsers are bringing up their support of WebSockets (and as the WebSockets protocol is being finalized: http://dev.w3.org/html5/websockets/).
    • +
  • Sockets are better sending data back and forth in an evented style rather than using a strict polling technique. Direct communication is just awesome…
    • +

So all in all, this is pretty cool. Making lights change on 5 different browsers on 2 different machines with one command is pretty awesome. Just a few things to wrap up and we’re golden to release the next stage in Kthxbye. Obviously this whole piece is pretty optional as you can always find another way to get your data back later on at your own time. This is just the way we tell the users, hey, your data is done. Come and get it!

+

A few things to note in the implementation details that are not well documented. 

+

Socket.IO is a bit tricky to get working with Rails. Rails handles GET requests for resources as well (unless you tell it otherwise) and Socket.IO has a number of paths hardwired (or set via JS variables) to look for the resources that it runs on (mostly other javascript files). Definitely follow the install instructions regarding getting recursive resources. For my current project, I had to include the socket.io directory in the public/ dir rather than the javascripts/ directory in public/. This can probably be gotten around but Rails complains that it cannot route some of the files. Its probably just an inclusion/paths issue on my side though. It just all started working once I moved it there, so I’ll leave it there until I start to move some of the custom stuff out of the Rails app and back into the plugin. io.setPath() apparently no longer exists which is pretty confusing, especially since the tutorial on the main page claims that’s the way to denote the socket.io libs if not serving it directly from Node.js).

+

Also, watch your proxies. Should be a given but I tried running this locally on two separate machines on our internal network and everything broke until I got the proxies working right… Stupid but a good reminder all the same.

+

The Flash socket protocol requires having port 843 open. Apparently that’s Adobe’s standard Flash port. I’m not sure this is working yet anyways, apparently the flash sockets are fairly sensitive…

+

Beyond that and some unsolved issue regarding why the flash sockets are still not working, this is a pretty intense project that has opened up a lot of new technology to me. Definitely will be using the redis/node.js combo a lot more in the future. 

+

UPDATE:

+

Rergarding the websockets issue, apparently, once you have all the paths correct to the .swf file that runs the websockets AND (very important and, up until the most latest release, undocumentedly), you MUST RUN YOUR NODE SERVER AS ROOT in order to get flashsocket support, which IE supports. Boom. All browsers now supported!

+

I do still have to add a symlink to my socket.io dir in my public/ dir in the rails app in order for GET requests to the flash file to work correctly. I may end up submitting a bug for that although it may be just something to work around with the way Rails serves files.

diff --git a/_posts/2010-09-30-month-in-music.html b/_posts/2010-09-30-month-in-music.html new file mode 100644 index 000000000..536873810 --- /dev/null +++ b/_posts/2010-09-30-month-in-music.html @@ -0,0 +1,12 @@ +--- +layout: post +title: Month in music +published: true +date: 2010-09-30 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1216931726/month-in-music +posterous_slug: post/1216931726/month-in-music +--- +

Been a long time:

+

[[posterous-content:jlcJDImjbjuHuBfDAzeB]]

+

[[posterous-content:fGtdeBCqsvwyorxAnrqv]]

diff --git a/_posts/2010-10-04-ie-gotcha-4-ie7-edition.html b/_posts/2010-10-04-ie-gotcha-4-ie7-edition.html new file mode 100644 index 000000000..41341e076 --- /dev/null +++ b/_posts/2010-10-04-ie-gotcha-4-ie7-edition.html @@ -0,0 +1,13 @@ +--- +layout: post +title: ! 'IE Gotcha #4: IE7 edition' +published: true +date: 2010-10-04 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1243153618/ie-gotcha-4-ie7-edition +posterous_slug: post/1243153618/ie-gotcha-4-ie7-edition +--- +

[[posterous-content:GftujlwFpHolACdCAumG]]In an ongoing attempt to alert the general public to all the downfalls you suffer when you use IE I continue the series: IE Gotchas. I haven’t hit any major ones in a while, but this is a weird one.

+

IE7 (and maybe before?) doesn’t give you the “JSON” object with which to manipulate json objects. Whaaat?

+

Simple solution is to require a conditional comment to include this file somewhere. Then you’re fine.

+

<!-- [if lte IE 7]> <script src="http://webscan.googlecode.com/svn-history/r80/trunk/ui/js/third-party/json2.min.js" type="text/javascript"></script> <![endif] -->

diff --git a/_posts/2010-10-05-kthxbye-1-1-0-rails-ified.html b/_posts/2010-10-05-kthxbye-1-1-0-rails-ified.html new file mode 100644 index 000000000..5787310dc --- /dev/null +++ b/_posts/2010-10-05-kthxbye-1-1-0-rails-ified.html @@ -0,0 +1,14 @@ +--- +layout: post +title: Kthxbye 1.1.0 - Rails-ified +published: true +date: 2010-10-05 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1249365580/kthxbye-1-1-0-rails-ified +posterous_slug: post/1249365580/kthxbye-1-1-0-rails-ified +--- +

Well 6 weeks worth of work on this baby and we now have a fully self-contained distributable project with slick Rails integration.

+

The initial release of Kthxbye focused on the machine behind the gem: the queuing processing of jobs. This release is more of a front-end and Rails compatibility release. This gem extracts out a lot of the built-in functionality that I had built around the gem into the primary application this gem was built to enhance. Mostly it involves the work I’ve been doing with Node.js and Socket.IO. It’s really simple code and if you want a look at the Node notification engine, it’s here. There’s a lot of new documentation added to the README file which ought to help you figure out how it works and what all needs to be installed. 

+

This is also my first experiment in Rails 3 > Railtie gemification. Railtie is really powerful engine an unfortunate few undocumented secrets (like that all generators apparently need to be named “<name>_generator.rb” in order to be picked up by Rails automatically). In any case, this gem is now super easy to integrate into your project that needs delayed job computation and callback notifications.

+

What’s next? Well, probably more features, a few enhancements, better reporting, and more slick ways to interact with Kthxbye.

+

Check it out now with “gem install kthxbye”. Enjoy all!

diff --git a/_posts/2010-10-06-internet-stop-being-so-intern.html b/_posts/2010-10-06-internet-stop-being-so-intern.html new file mode 100644 index 000000000..a909a20bb --- /dev/null +++ b/_posts/2010-10-06-internet-stop-being-so-intern.html @@ -0,0 +1,13 @@ +--- +layout: post +title: ! 'GIANT ROBOTS SMASHING INTO OTHER GIANT ROBOTS: Internet: Stop being so Internety' +published: true +date: 2010-10-06 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1256508776/internet-stop-being-so-internety +posterous_slug: post/1256508776/internet-stop-being-so-internety +--- +http://robots.thoughtbot.com/post/1256508776/internet-stop-being-so-internety

thoughtbot-giantrobots:

+
+

Recently Gap redesigned their logo and I’m bothered by how we, the design community, are often responding. Dribbble and Twitter are great tools. I use both of them frequently. They are also vessels for people to freely speak their mind. Unfortunately, it’s too often used to degrade and belittle…

+
diff --git a/_posts/2010-10-09-new-project-in-the-works.html b/_posts/2010-10-09-new-project-in-the-works.html new file mode 100644 index 000000000..df75a3b05 --- /dev/null +++ b/_posts/2010-10-09-new-project-in-the-works.html @@ -0,0 +1,12 @@ +--- +layout: post +title: New Project in the Works +published: true +date: 2010-10-09 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1274087562/new-project-in-the-works +posterous_slug: post/1274087562/new-project-in-the-works +--- +

More design focused this time. Sneak peak:

+ +

[[posterous-content:iDFApilBvGqzFpmczlxs]]

diff --git a/_posts/2010-10-25-what-i-have-decided.html b/_posts/2010-10-25-what-i-have-decided.html new file mode 100644 index 000000000..57faeb556 --- /dev/null +++ b/_posts/2010-10-25-what-i-have-decided.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-10-25 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1398983003/what-i-have-decided +posterous_slug: post/1398983003/what-i-have-decided +--- +

[[posterous-content:xDfvszmIrCwpizGgeIJw]]

What I have decided.

diff --git a/_posts/2010-10-27-rails-3-gotcha.html b/_posts/2010-10-27-rails-3-gotcha.html new file mode 100644 index 000000000..c4fb1c1bd --- /dev/null +++ b/_posts/2010-10-27-rails-3-gotcha.html @@ -0,0 +1,14 @@ +--- +layout: post +title: Rails 3 Gotcha +published: true +date: 2010-10-27 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/1412661181/rails-3-gotcha +posterous_slug: post/1412661181/rails-3-gotcha +--- +

This is a new one. Ran into this whilst deploying a new server tonight. Rails 3 defaults to use XSendfile to force the HTTP server (Apache/Nginx) to handle all static file sends.

+

So here’s the deal, this fun little Apache module is buried deep in the pages of the interwebs. It’s not super well known or documented.

+

So for my everyone’s peace of mind, the docs and source can be found here: https://tn123.org/mod_xsendfile/

+

The trouble is what happens when you try to use send_file, is Rails will handle the request perfectly, down to renaming the very empty file you just sent the user when XSendfile is not installed, enabled or properly configured. Its not hard, its just not easily documented or discovered. So after almost an hour of trying to remember how I fixed this, I ran into this article, which details the exact same problem/output I saw.

+

Updated, configured, fixed. Thanks internets.

diff --git a/_posts/2010-12-26-include-vs-extend-in.html b/_posts/2010-12-26-include-vs-extend-in.html new file mode 100644 index 000000000..26f03e495 --- /dev/null +++ b/_posts/2010-12-26-include-vs-extend-in.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Difference in adding class and instance methods +published: true +date: 2010-12-26 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/blog/archives/2009/05/15/include-vs-extend-in-ruby +posterous_slug: blog/archives/2009/05/15/include-vs-extend-in-ruby +--- +http://railstips.org/blog/archives/2009/05/15/include-vs-extend-in-ruby/ diff --git a/_posts/2010-12-26-job-new-company-salescrunch.html b/_posts/2010-12-26-job-new-company-salescrunch.html new file mode 100644 index 000000000..21aa1cca5 --- /dev/null +++ b/_posts/2010-12-26-job-new-company-salescrunch.html @@ -0,0 +1,24 @@ +--- +layout: post +title: Job.new(:company => "SalesCrunch") +published: true +date: 2010-12-26 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/2473837668/job-new-company-salescrunch +posterous_slug: post/2473837668/job-new-company-salescrunch +--- +

So long time, and here’s why.

+ +

At the tail end of this year, I’m starting in with a brand new company, SalesCrunch. It’s an exciting new opportunity for me and I’m really pumped to be working with a group of talented and experienced individuals who aren’t working around Oracle (sorry Hargray).

+ +

SalesCrunch’s site gives some details and a cool video about what the project it, but the developer lowdown is, we’re making a crazy awesome product with a lot of really cool technologies on the back end. I’m gonna have lots to blog about in the coming days, when I find the time. Things have been really fast, with a few major deadlines looming (soft-launch was during my first week! No time to start like during the launch cycle!) but its keeping me interested and busy.

+ +

Three things make this job a huge step above where I had been:

+ +
    +
  1. Pairs programming
    2. +
  2. TDD
    3. +
  3. Agile development processes.
    4. +

These are awesome for a number of reasons. Pairs got me up to speed and made me a productive, contributing member to the project on the first day, and TDD is making sure all the code I contribute integrates and works. I’ve never done either before and I’m really pumped to be working with both. Agile development is pretty sweet too, because it’s giving me a small subset of a larger goal and making me feel productive on a day-to-day basis. A lot of previous projects were over-arching “We need X done by Y”. Now its, “Lets get these pieces done today, over the week, to work toward, X.” I can’t say enough for both already.

+ +

There will be plenty more to come. Stay tuned on this brand new journey, and definitely check out SalesCrunch and sign up for an invite! Ping me in the comments if you want one and I’ll see what I can do.

diff --git a/_posts/2010-12-26-rock-stars-of-code-when-a-cul.html b/_posts/2010-12-26-rock-stars-of-code-when-a-cul.html new file mode 100644 index 000000000..80158fdbf --- /dev/null +++ b/_posts/2010-12-26-rock-stars-of-code-when-a-cul.html @@ -0,0 +1,53 @@ +--- +layout: post +title: Rock Stars of Code - When a Culture Takes a Turn for the Worse +published: true +date: 2010-12-26 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/2474143652/rock-stars-of-code-when-a-culture-takes-a-turn-for +posterous_slug: post/2474143652/rock-stars-of-code-when-a-culture-takes-a-turn-for +--- +

Note, I started this a couple months back and am just now getting the time to finish. Dates/times may be off now.

+ +

Finally a bit of a break with some time to write. Busy and in the middle of switching jobs while training up new people for my old job. These thoughts come more out of RubyConf as I flew back home a few weeks ago.

+ +

There is an unusual trend in the developer culture that has not gone unnoticed by many people. It’s the idea of “rock star coders”. You know what I mean. Those people who we all hold in high regard because of the body of work or the high-visibility of the projects they have worked on. Its an unusual phenomenon which, as far as I have seen, only two developer circles suffer from: iOS developers and Ruby developers. Think about it. How many “rock star” PHP developers can you name? How many people would you put in the Java hall of fame? Okay, how about Ruby?

+ +

Here’s just a few off the top:

+ +
    +
  1. Chris Wanstrath
    2. +
  2. Mike Perham
    3. +
  3. Ryan Tomayko
    4. +
  4. Nick Quaranto
    5. +
  5. DHH
    6. +

Or Mac/iOS?

+ +
    +
  1. Cabel Sasser
    2. +
  2. Matt Gemmell
    3. +

And I’m sure we could all name many more in both categories. Now some would ask, “Why is this a bad thing? It’s awesome our languages/industries/social circles have heros! We’re usually the uncool, nerdy class. Now we can be awesome!” While I certainly approve of the positive press developers are starting to get, the whole idea of coder rock-stars strikes me as a step back. I certainly like that we can have high-visibility people who give us inspiration and lots of quality code, but when we switch to idolization of anyone (not just developers), we are doomed to sit in their shadow and so stunt our own creativity. I have realized that the time I have spent admiring individuals and their code has meant so much less time spent developing my own coding style and technique. I’ve created much less because I’ve sat in awe of the “Big Names” in the Ruby/Rails community. I started thinking about working hard, learning the Rails core in order to become a core contributor so that I could be seen as a “Big Name” in my own right in the community. How awesome would that be?!

+ +

As I bounced from talk to talk at RubyConf, I realized that, while I was seeing a lot of cool stuff, I began to get in the mindset that I’ll never make anything as cool as these guys have and will. And when you end up there, motivation to write new and innovative goes out the window.

+ +

Which sucks. For you and for the community you are apart of.

+ +

What would be far better for us is to use what is being done to fuel our own ideas and creativity, not store up awe and reverence for earthly beings, who are making strides by looking ahead and not around.

+ +

=== Waxing philosophical

+ +

Society in general is very competition-driven. We want to 1-up the Jones’. We want to beat that 14yo. at Call of Duty: Black Ops. We want to write code that is cutting-edge and used by millions. It’s a natural attitude.

+ +

It’s also wrong.

+ +

We tend to idolize those who stand out as above and beyond those in their own class. Sports is a huge example of this. That’s really the point of sports. But then it starts to move into other areas of life. Like work. And usually this creates a less-than-healthy environment/mindset for working.

+ +

Competition is an odd fuel for productivity. We often think competition in the workplace should help build products or more fascinating code. Yet when we compete, someone else inevitably ends up on the bottom. And that’s bad for morale.

+ +

And bad morale is bad for everyone.

+ +

So a word to the community I am a part of and those I know by extension: Put away the idolization of developers (or anything else). Do your job. Create, innovate, do something new. Do something old if it means learning something new. Reinventing the wheel teaches you how to make wheels. Then you can take that knowledge and do something with wheels that no one has done before a lot better than someone who’s never made a wheel.

+ +

Most of all, work for the community and the benefit of others, not for rock-stardom. In the end, it benefits us all more than making monumental changes to high-profile projects. And I’m not just talking about code.

+ +

Do it in your community, whether that’s a workplace, a church, a school or your neighborhood. So much good could be done by working for the benefit of others rather than the grandifying of self.

diff --git a/_posts/2010-12-27-rock-stars-of-code-2-a-side-e.html b/_posts/2010-12-27-rock-stars-of-code-2-a-side-e.html new file mode 100644 index 000000000..63a7bf41d --- /dev/null +++ b/_posts/2010-12-27-rock-stars-of-code-2-a-side-e.html @@ -0,0 +1,12 @@ +--- +layout: post +title: Rock Stars of Code 2 - A Side Effect +published: true +date: 2010-12-27 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/2476560398/rock-stars-of-code-2-a-side-effect +posterous_slug: post/2476560398/rock-stars-of-code-2-a-side-effect +--- +

Reflecting on the last post, one other thing came to mind regarding rock-stardom. In trying to interact with many of the “head’s of state” at RubyConf (Aaron Paterson, Obie Fernandez, etc.), while I was greeted cordially, it often goes no further than that. Part of the dangers of stardom in any circle is that you start to believe you’re a big deal. You become too big to be bothered by the common coder. You make the money, you have the groupies. Now you are the elite, and only those who share that elite may get a word in edgewise. Part of what I want to see is those that have the most to contribute to code, also have the most to give to others. The most beneficial conversations I had were with people who had nothing to gain from a conversation other than friendship. Developers who loved doing what they do and want to share that passion for a platform or a community with others. Those who are working all hours to enable others to go the extra mile and be great. Not for their own glory, but for the sake of community. Those are the people I love to meet and interact with and work along side.

+ +

Be those people. You do us all proud.

diff --git a/_posts/2010-12-31-make-it-not-as-readable.html b/_posts/2010-12-31-make-it-not-as-readable.html new file mode 100644 index 000000000..5a3554fd4 --- /dev/null +++ b/_posts/2010-12-31-make-it-not-as-readable.html @@ -0,0 +1,11 @@ +--- +layout: post +title: Untitled +published: true +date: 2010-12-31 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/2535470499/make-it-not-as-readable +posterous_slug: post/2535470499/make-it-not-as-readable +--- +
Make it not as readable.
+ diff --git a/_posts/2011-01-25-three-weeks-of.html b/_posts/2011-01-25-three-weeks-of.html new file mode 100644 index 000000000..3e83fd44b --- /dev/null +++ b/_posts/2011-01-25-three-weeks-of.html @@ -0,0 +1,42 @@ +--- +layout: post +title: ! 'Three Weeks of... ' +published: true +date: 2011-01-25 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/2925231760/three-weeks-of +posterous_slug: post/2925231760/three-weeks-of +--- +

Today marks two and a half weeks into a major enhancement to CrunchConnect, one of the biggest pieces of the products SalesCrunch is putting forward.

+ +

First let me preface this by saying that TJ Singleton is a god at JavaScript and pretty much anything else. He’s almost single-handedly written the NodeJS backend and Redis/Node ORM for this project and on top of that given me all the right answers to what I so poorly understand about our adventures into JS-only server-client code. Also David Sommers and Gabi Moore have done an amazing job putting together the amazing interface design we have and its going to blow any other online, community presentation webapp you’ve seen away). For realz.

+ +

Second, this product is going to be awesome. If you have anything to do with sales or online presentations to clients, etc., signup for the beta RIGHT NOW. Feel free to ask for invites in the comments and I’ll do what I can to help you out!

+ +

Okay so on to the layers.

+ +

For the past 2.5 weeks we’ve been hammering away at a slick JS system that is comprised of a NodeJS server that has a custom, built-in Redis ORM for managing users, meetings, and presentations (including the states of each of those objects). Its been such a trip and I’ve learned so much about what JS can do and is capable of, I figured I’d chronicle some of the trip here.

+ +

So I’m a Ruby developer through and through, and something TJ commented early on about was that Rails has a bad way of bending your mind to a singular way of thinking about MVC. Rails handles the MVC pattern extremely well and separates it out really cleanly. But MVC is not Rails-specific, nor is the way Rails does MVC the only way to do a MVC pattern.

+ +

In steps JavaScript without a lot of form, figure or complication and a crapload of speed. We use jQuery for interface interaction, and Socket.IO and Redis Pub/Sub to do a lot of our communication with the Node server. From there, we’re pretty much on our own as far as existing technologies. We’re using Jasmine on the client side and Vows on the server side to test, which has helped or code consistency and reliability and torn our productivity to shreds (at times).

+ +

The Implementation

+ +

TJ and I started paring on the ORM. He wrote, I pretty much watched, checked and gave color-commentary as we moved along. Pretty simple and modeled amazingly closely to ActiveRecord’s way of things. Anything from has_many to has_and_belongs_to_many, we have it. What’s even cooler is it backends to Redis, so its pee-your-pants fast. This was all built in Node (via CoffeeScript I might add). We talked about open-sourcing it, but we’ll see where that goes.

+ +

From here TJ has handled most of the Node implementation while I’ve worked on client side in step with him so that as different parts of the protocol came together, I could send events, he could respond to them and we could both test each other’s code as we wrote it. I also worked at the same time to integrate the Node responders with the interface as it was built out so that we could see things happening like they would during a live meeting. Seeing stuff happen on an interface rather than just in a command-line interface can make a world of difference on your morale, especially when you’re writing Javascript.

+ +

Our server/client setup is very much state-driven. Node is built around an evented paradigm, so our implementation of a server-client system is almost entirely evented. We send requests, the server sends response events back or schedules event jobs (via Resque) to update data in our DBMS database. So all in all, if nothing is happening on the client, neither is anything going on on the server. Saves us a ton of bandwidth, memory and CPU cycles.

+ +

As I mentioned, we are using Socket.IO as our primary communication transport, which means signals to and from the client/server happen in milliseconds after being triggered. This is important as this system handles everything from our chat service (a system which generally allows some latency) to our user status and presence UI control, to our conference calling UI updates and controls which needs to transmit to hundreds of people at the same time. Our protocol by which we pass messages is also extremely small. We’ve made it very generic and kept data passed back and forth to a minimum using JSON in a very basic form to keep data across the wire low.

+ +

Javascript

+ +

Javascript is a weird beast. It mimics both functional and OO langages. And the object design paradigms change radically from what I have been used to in Ruby/Rails. Moreover, all the “magic” Rails handles for you is gone when dealing with UI or request routing. Beyond that, we’re integrating the JS client with our Rails app so there’s a fair amount of Ajax data getting and setting going on. I’ve never written much .bind/.trigger stuff before. But our whole system is built around triggering the client based on events bound to the protocol we wrote, in large part based around a simplified set of the the IRC protocol. We use callbacks a lot to simplify, “genericize” and asynchronize the general operation of the communication lifecycle. It makes it harder to track how a request flows through the system, but makes it much easier to maintain individual parts of the request cycle and makes it flexible in extending it to respond to new requests as we expand the feature-set of the product. This is what the Rails core team worked towards in Rails 3 and is a pattern we worked to follow in our own way.

+ +

Summarize

+ +

Writing something like a chat app isn’t that hard. But when you actually need to track users in specific rooms, integrate that with conference calling, connectivity/presence tracking, admin control over users and presentations and rooms, it just gets crazy.

+ +

But we’re doing it. And its going to be awesome, not only in the incredible backend we’re working on, but in the beautiful interface that we’re supporting and integrating into. SalesCrunch is an awesome company with amazing minds at work, and CrunchConnect is going to be one hell of a product.

diff --git a/_posts/2011-03-08-an-open-letter-from-our-vp-of.html b/_posts/2011-03-08-an-open-letter-from-our-vp-of.html new file mode 100644 index 000000000..0e3302ce6 --- /dev/null +++ b/_posts/2011-03-08-an-open-letter-from-our-vp-of.html @@ -0,0 +1,22 @@ +--- +layout: post +title: An open letter from our VP of Engineering +published: true +date: 2011-03-08 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/3723792894/an-open-letter-from-our-vp-of-engineering +posterous_slug: post/3723792894/an-open-letter-from-our-vp-of-engineering +--- +

We’re a Rails application hosted on Heroku. As the demand for access to our system increased, our ability to auto-scale failed.

+ +

Our original auto-scaling technique was wired into rack to monitor queue lengths at timed intervals. The problem with our code was it wasn’t that smart with scaling up and then back down. It would scale up real fast for workers under very little load and then stay high for some time before coming back down because it was fairly conservative. It didn’t work well with dynos at all, there was usually a long delay to warm up new dynos when requests started queuing. Mainly because of a mutex problem with who owns the rights to change the dyno count?

+ +

HeroScale was then released and we switched to that for some time until… they couldn’t process custom URLs. Our URL path to the server isn’t at heroku.com as most applications can be found. We have a little bit of rack middleware to verify the URL of our site since we will be introducing 3rd level domain names for personalized sites at some point. The middleware interferes with HeroScale’s ability to adjust settings and so we had to dump HeroScale a few weeks ago. Additionally, HeroScale is based on the assumption that it can hit your web site’s status page at /heroscale/status. Unfortunately for us, that URL was offline due to the traffic spike and DDoS like characteristics of being Slashdotted. So even if we had it running, it wouldn’t have worked.

+ +

The solution?

+ +

We’re researching auto-scaling via resque-scheduler and resque workers, which if works, we’ll release on github. When we first developed our custom solution, we were still on DelayedJob and we didn’t have a scheduling option - only daily and hourly crons. By switching to a worker based solution, we hope to build out a god-like monitor for the cloud running on the cloud. If the site becomes under load or simply unavailable, crank it up!

+ +

Without hosting ourselves, the ability to monitor other people’s resources is an interesting problem. You have the power to scale, which is a benefit of the cloud, but only to the extent that the vendor’s tools allows. It would be nice if Heroku did provide auto-scaling, and I’m interested in DuoStack which will provide auto-scaling. Once their services reach a certain level of maturity, we may switch providers but for now - we’ll have to roll our own, keep monitoring everything, and tweak as we go.

+ +

A normal day for any geek in a startup: developer + sysadmin == lots to do!

diff --git a/_posts/2011-03-28-read-the-damn-code-rtdc-episo.html b/_posts/2011-03-28-read-the-damn-code-rtdc-episo.html new file mode 100644 index 000000000..373efe020 --- /dev/null +++ b/_posts/2011-03-28-read-the-damn-code-rtdc-episo.html @@ -0,0 +1,24 @@ +--- +layout: post +title: ! 'Read the Damn Code (RtDC) - Episode 1: Union vs. Array#uniq!' +published: true +date: 2011-03-28 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/4163548174/read-the-damn-code-rtdc-episode-1-union-vs +posterous_slug: post/4163548174/read-the-damn-code-rtdc-episode-1-union-vs +--- +

So as part of a quick request to modify our very meta-stats system, I got a first look at the code that runs the stats generation/reporting for our site’s activities. Can I say again, its awesomely meta? (Props to @tjsingleton).

+

So basically we setup custom hashes that grab data from the DB that contain common stats that interest us and our customers, like how many emails got sent on a given day, how many presentations were given, how many invites were sent or accepted. Pretty easy stuff. The magic happens in translating it from the data into graph-interpretable data, which, again, is not too complicated. It just means formatting our hashes into something the graphing engine can read and render.

+

So my quick request was just to change the order of a few column headers in the stats reporting table. Easy? Well. So everything gets thrown into a single hash unless we’re calculating the stats from previously gathered stats. For example, live presentations is the difference between total presentations and emailed presentations. Ruby lets you do simple Array1 - Array2 = (x: x ∈ Array1 and x ∉ Array2). Simple difference right? Cool. But these special calculated hashes are stored in a separate array, which I found out, are then combined via Array#| (union) later on. Which creates an order problem. Since union has no means of ordering elements, it just appends the additional elements onto the end of the returned array. Which is fine unless we want to keep the order in which we defined the keys (in say some .rb or .yaml definition file).

+

So what to do? Simple. Store the keys in their own array aside from the hashes the data is being appended onto and run uniq! when we request the value of the keys to ensure we get all unique values. A good idea?

+

I found out a few things about a better way (although someone else may have the “best way”) to do some of these things. First of all, don’t do

+

keys << key unless keys.include? key

+

Why? include? has to loop through the entire array to figure out if your element exists. The bigger your key array gets, the longer that takes. Return time for that hash will be incredibly fast though, as even uniq! is slower than simply returning an already unique array. I wanted to graph the tradeoff point that you should do this, but I figure the description alone is obvious enough.

+

Second, there’s danger in keeping a key list separate from your actual data list. List syncing could be tricky to keep up with. Fortunately, this is an immutable list (for now). Once you load up the page, the stats are there to stay. This brings into question (why not use the include? code above if your list is small enough and you only load once? Webscale bitches.

+

uniq! is really nice in that it converts the array to a hash to do the uniqueness check. This makes the data type handle the uniquifying aspect of making an array contain unique values. Brilliant? Yes. But digging deeper, how does this conversion work? Here’s the code for array.c. It loops through each value of the array and adds it as a key to a disposable hash. Wait, looping each value again? How is this better? It’s not so much about that it loops as it is when it loops. We run the uniq! only when the data is requested. This means a few things. 1) Once we uniq!, our array is updated and any further uniq! will have a smaller set of data to operate on. 2) We’re running this when the data is requested, rather than on every add. Since we don’t care about the data until we ask for it, we can save data-load operations by waiting until the last minute to ensure our data will show correctly.

+

Finally, what does the union do (Array#|)? Not surprisingly, pretty much the same thing as uniq! (codez). It creates a disposable hash from the array, and then tries to add any non-existing keys to the hash, using the hash as the uniqueness enforcer. The main difference is that union has to do this twice for the two arrays it is union-ing. Plus it creates a lot of other arrays in memory to hold the intermediate results. The main problem with using union in my case is that it appends new elements to the end, rather than giving us a way to keep order between two arrays, which is what we needed in the first place.

+

So, there you go. The answer is always in the code. Not a necessary look at the details, but one inspired to understand how a language works and how can we work faster.

+

A few nearly useless benchmarks too!

+

+

TL;DR

+

Do your homework. Read the code. Think about the maths (and the Matz). It’s always worth it, if not in performance, then in understanding.

diff --git a/_posts/2011-04-02-the-challenges-of-working-remotely.html b/_posts/2011-04-02-the-challenges-of-working-remotely.html new file mode 100644 index 000000000..abdf94e53 --- /dev/null +++ b/_posts/2011-04-02-the-challenges-of-working-remotely.html @@ -0,0 +1,10 @@ +--- +layout: post +title: The Challenges of Working Remotely +published: true +date: 2011-04-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/article/the-challenges-of-working-remotely +posterous_slug: article/the-challenges-of-working-remotely +--- +http://cognition.happycog.com/article/the-challenges-of-working-remotely diff --git a/_posts/2011-04-04-possessive-tendencies.html b/_posts/2011-04-04-possessive-tendencies.html new file mode 100644 index 000000000..2a984a4be --- /dev/null +++ b/_posts/2011-04-04-possessive-tendencies.html @@ -0,0 +1,18 @@ +--- +layout: post +title: Possessive Tendencies +published: true +date: 2011-04-04 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/4341996297/possessive-tendencies +posterous_slug: post/4341996297/possessive-tendencies +--- +

I have recently discovered an interesting tendency in my thinking patterns about code. Before the past few months, I would have said that I am not a terribly possessive person. I like OSS and collaborating on projects. But for the first time in my professional career, I am working on a project on which I am collaborating with other developers. As I have done so, I have found a strange feeling of pain, or discomfort with other developers changing my code. I worked so hard to get this to work, to make that happen, to optimize that runtime. And then someone goes and changes it.

+ +

This is bad, mmk?

+ +

Being possessing with code leaves no room for other people to work. It also builds feelings of animosity within a team. It can be easy to feel judged, as if your code isn’t good enough and needed to be redone. Most of the time, rewriting code is good. First runs are often just with the goal of “make it work” or “get tests to pass”. Second (or third) pass is for refactors and optimizations. What’s more, I’m still in the mid-level developer category. I have lots to learn and I’m fortunate to be on a team with a senior-level developer (if these classifications actually mean anything) who has more experience and a mind with a good grasp of writing code for flexibility (webscale, son). I’m learning a lot from how he rewrites my code or by pairing with him over changes assigned to me.

+ +

Don’t possess code. Get other people to look it over. Let it change easily. Be open to criticism or change rejections. It only makes the code better and give you more insight into how to write better code.

+ +

Yay community.

diff --git a/_posts/2011-04-06-salesschool-nyu.html b/_posts/2011-04-06-salesschool-nyu.html new file mode 100644 index 000000000..9b6845b86 --- /dev/null +++ b/_posts/2011-04-06-salesschool-nyu.html @@ -0,0 +1,197 @@ +--- +layout: post +title: ! 'SalesSchool: NYU' +published: true +date: 2011-04-06 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/4390636852/salesschool-nyu +posterous_slug: post/4390636852/salesschool-nyu +--- +

The following are my notes from a presentation given by Jan Schultink, CEO of Idea Transplant. He came and gave a talk as part of our SalesSchool program to give relevant content to sales people, in particular for this evening, those who are seeking VC funding.

+ +

Also note, we have another event tonight, again at NYU. More great material focused around much of the same ideas, but more geared towards sales in general, rather than VC pitches.

+ +

The video will be up on the site some point in the near future, but for now, here’s my take on the presentation.

+ +

How to Design Stunning Sales Presentations

+ +

Images

+ +

No bullet points: means you must construct, audience is faster at reading then you will be at explaining through points.

+ +

instead…

+ +

Images are emotional shortcuts to make points.
+Packages information much faster. Less need to explain.
+Brings emotion to connect to information and boost memory.

+ +

Daunting images for backgrounds will burn visual trademark to the information.
+Gives big memory recall triggers.

+ +

source: iStockPhoto.
+warning: easy to find a person who is duplicated many times. be unique.
+develop an eye for picking.

+ +

source: flickr CC
+warning: quality. be sure to use creative commons, w/ attribution.

+ +

Watch proportions.
+Ban clip art.
+Use advertisements as examples or tutorials for presentations. Good/valuable trends show up here.

+ +

Leave much whitespace.
+No slide limit, use a time limit instead.

+ +

symmetry is bad. use the rule of thirds.

+ +

align. poor alignment distracts.

+ +

Sequence - momentup.

+ +

Start left and lower, up and to the right gives a feeling of energy and positive movement.

+ +

Easy on the animations. Keeps it simple and professional.

+ +

Numbers: simplify and round. No need for hyper-accuracy.

+ +

Charts. Don’t do small axis labels, don’t need 3D graphics. Simplify, show single trends.

+ +

Use blank slides in order to bring attention back to presenter.

+ +

Hygiene - general look and feel.

+ +

No comic sans.
+No odd colors.
+Use professional email addresses with company domain name

+ +

Colors: Adobe Kuler.
+Be consistent across presentation. Same colors for all slides.

+ +

Give content space. Don’t do a lot of repeating elements.
+Use colors that fit with your corporate identity.

+ +

Stories.

+ +

Do not use an agenda slide. Repetition does not help. Only gives a feeling of being forced.

+ +

Presentations are not references. Not business plan driven. Its more driven by trying to find holes.

+ +

Using very short “stories”/information snippets will cause others to want the information, to start to try to find and fill in the details.

+ +

Mind memorizes by placing information in spaces (like a picture of a whiteboard)

+ +

Think product placement within a story.

+ +

Don’t look at powerpoint, use a sheet of paper or listen to yourself to build a presentation.
+Will give you a roadmap to build on and find the important points.

+ +

Take time to build presentation. 2 weeks for development + 2-3 days to build slides.

+ +

Content - for the VC

+ +

Team intro is helpful at beginning, doesn’t have to be on slides.

+ +

Presentation is not just the slides. Also many other inputs. Such as:

+ +

Gut conviction + Head conviction.
+Must pull and convince both.

+ +

YOU. You carry the deck along. The body language speaks.

+ +

Convince me that:

+ +
    +
  • People will buy
    • +
  • You can sell it
    • +

Build trust through honesty.
+No: “traffic is thru the roof” -or- “we have no competitors”
+Feel free to voice uncertainty, with conviction you are still right.

+ +

Types of conversations:

+ +
    +
  • 10 second conf chat + +
      +
    • Use sentences, not massive amounts of buzzwords.
      • +
    • Made to Stick (book)
      • +
    • Don’t leave people with just a random tune stuck in their heads.
      • +
    • Mission statements have no bearing on those who have no context to work in
      • +
    • Go with specific issues that may just be a single part of the picture
      • +
    +
    • +
  • 30 minute pitch + +
      +
    • TED is 20 minutes.
      • +
    • No need to fill all the time.
      • +
    • Be concise and use remaining time for dialog
      • +
    • Don’t preach to the converted. Ideas already taking off don’t need to be given more value.
      • +
    • Don’t slap all your ideas together into a frankenstein deck.
      • +
    • Pick up signals presentees are sending as responses to your info + +
        +
      • Address “elephant in the room issues” early to remove distractions.
        • +
      • Will help focus the investor.
        • +
      +
      • +
    • Focus on: What are you doing currently? Even if its a rougher overall picture.
      • +
    • Pitch the problem rather than the features of the solution.
      • +
    • Use visuals to show problems so that the solution becomes apparent by itself.
      • +
    • Maintain momentum. Demo of products will kill this more than anything. + +
        +
      • Maybe use screenshots instead.
        • +
      • Let them play afterwards.
        • +
      +
      • +
    • “If you can’t explain it to a 6 year old, you don’t understand it yourself” - Einstein
      • +
    • Show very short term plan in great detail. Show long term in more ballpark terms.
      • +
    • Go simpler the greater the lengths of time
      • +
    • Actually justify long term numbers. Plan.
      • +
    • Be consistent.
      • +
    • Framework structures are boring. Recreate as your own using basic/simple diagrams.
      • +
    • Highlight a complimentary team that works together well.
      • +
    +
    • +
  • 15 minute version + +
      +
    • Print out, don’t staple. Easy to move around within a presentation. Use paper. More flexible
      • +
    +
    • +
  • Board Meeting + +
      +
    • will not depend entirely on deck
      • +
    • move off to excel or detailed data transmission tools.
      • +
    +
    • +

For potential “pre deck screenings” either give information as part of the slide, write out the message in a side column, or create two decks.

+ +

Webinars

+ +

Get technology set up correctly
+Page reference for fast switching (easy to go back or move around within deck)
+Avoid 5 second charts. Charts need to be focus points.
+Make it about fewer slide transitions.
+Build up with pointers.
+Explain why you are silent (drinking water, etc).

+ +

Learn

+ +

Presentation Zen - Garr Reynolds
Brain Rules - John Medina

+ +

Meeting notes in public domain:
Mark Suster, Fred Wilson
+Gain insight into successful presentations. Some VCs give pointers to what they liked.

+ +

One more thing.

+ +

Practice!!! +Rumor has it, Steve Jobs takes 3 working days to practice.
+Must be on top of/know totality of content in order to be spontaneous.

+ +

Always make point early. Don’t get cut off.

+ +

What slide stays on the longest? The last one. Usually says “Q&A”…
+Should be something that you discussed to help burn into minds as they sit doing something else.

diff --git a/_posts/2011-04-08-a-new-thing-starting-at-sales.html b/_posts/2011-04-08-a-new-thing-starting-at-sales.html new file mode 100644 index 000000000..674f19825 --- /dev/null +++ b/_posts/2011-04-08-a-new-thing-starting-at-sales.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2011-04-08 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/4446525388/a-new-thing-starting-at-salescrunch-we-have-a +posterous_slug: post/4446525388/a-new-thing-starting-at-salescrunch-we-have-a +--- +

[[posterous-content:zIfAGwGEtCmBfagxyynv]]

A new thing starting at SalesCrunch. We have a mascot, and his name is Steve.

diff --git a/_posts/2011-04-12-on-meetings-from-a-meeting-so.html b/_posts/2011-04-12-on-meetings-from-a-meeting-so.html new file mode 100644 index 000000000..2dfb555b5 --- /dev/null +++ b/_posts/2011-04-12-on-meetings-from-a-meeting-so.html @@ -0,0 +1,10 @@ +--- +layout: post +title: On meetings... from a meeting software company. +published: true +date: 2011-04-12 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/4560965978/on-meetings-from-a-meeting-software-company +posterous_slug: post/4560965978/on-meetings-from-a-meeting-software-company +--- + diff --git a/_posts/2011-04-27-easter.html b/_posts/2011-04-27-easter.html new file mode 100644 index 000000000..d42b99736 --- /dev/null +++ b/_posts/2011-04-27-easter.html @@ -0,0 +1,18 @@ +--- +layout: post +title: Easter +published: true +date: 2011-04-27 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/4974756109/easter +posterous_slug: post/4974756109/easter +--- +

This is not my traditional venue for sharing these kinds of thoughts or links, but this is something near and dear to my heart that anyone who knows me ought to hear.

+ +

So many of the people I follow or know within technical or developer circles distain, abhor, make fun or light of what and Who Easter is about. This is the day that Jesus Christ rose from the grave, demonstrating that all he said was true.

+ +

John 20:30-31 says: “The disciples saw Jesus do many other miraculous signs in addition to the ones recorded in this book. But these are written so that you may continue to believe that Jesus is the Messiah, the Son of God, and that by believing in him you will have life by the power of his name.

+ +

You can disagree with me as you like, but I believe the message of this video with all my heart. Watch with an open mind and heart. I pray it changes your life.

+ +

Mars Hill, Mark Driscoll - Easter 2011

diff --git a/_posts/2011-05-10-lessons-learned.html b/_posts/2011-05-10-lessons-learned.html new file mode 100644 index 000000000..ca832dd10 --- /dev/null +++ b/_posts/2011-05-10-lessons-learned.html @@ -0,0 +1,27 @@ +--- +layout: post +title: Lessons Learned +published: true +date: 2011-05-10 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/5361931587/lessons-learned +posterous_slug: post/5361931587/lessons-learned +--- +

So what have I learned by venturing out in the unexplored labyrinth of Javascript application development?

+ +

Its not as hard as you think

+ +

As developers, whether formally trained via the architects of higher education or simply learned through experience and mentorship, we are natural-born learners. We were taught to figure things out by experimentation and exploration of available knowledge. What’s even better is most developer communities are also eager to spread knowledge. Maybe its the fact that we like learning from others that we also desire to be learned from. But the sheer amount of information available is simply staggering. I’ll put some of my most referenced articles as part of the recent foray into JS.

+ +

There have been many more on this list that have come across my browser as I search for stuff, but these have been influential and oft returned to as this project has progressed.

diff --git a/_posts/2011-05-10-possessive-tendencies.html b/_posts/2011-05-10-possessive-tendencies.html new file mode 100644 index 000000000..951e910f1 --- /dev/null +++ b/_posts/2011-05-10-possessive-tendencies.html @@ -0,0 +1,20 @@ +--- +layout: post +title: Possessive Tendencies +published: true +date: 2011-05-10 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/5361916738/possessive-tendencies +posterous_slug: post/5361916738/possessive-tendencies +--- +

Possessive Tendencies

+ +

I have recently discovered an interesting tendency in my thinking patterns about code. Before the past few months, I would have said that I am not a terribly possessive person. I like OSS and collaborating on projects. But for the first time in my professional career, I am working on a project on which I have to collaborate with other developers. As I have done so, I have found a strange feeling of pain, or discomfort with other developers touching my code. I worked so hard to get this to work, to make that happen, to optimize that runtime. And then someone goes and changes it.

+ +

This is bad, mmk?

+ +

Being possessing with code leaves no room for other people to work. It also builds feelings of animosity within a team. It can be easy to feel judged, as if your code isn’t good enough and needed to be redone. Most of the time, rewriting code is good. First runs are often just with the goal of “make it work” or “get tests to pass”. Second (or third) pass is for refactors and optimizations. What’s more, I’m still in the mid-level developer category. I have lots to learn and I’m fortunate to be on a team with a senior-level developer (if these classifications actually mean anything) who has more experience and a mind with a good grasp of writing code for flexibility (webscale, son). I’m learning a lot from how he rewrites my code or by pairing with him over changes assigned to me.

+ +

Don’t possess code. Get other people to look it over. Let it change easily. Be open to criticism or change rejections. It only makes the code better and give you more insight into how to write better code.

+ +

Yay community.

diff --git a/_posts/2011-05-10-top-skype-owner-microsoft-twe.html b/_posts/2011-05-10-top-skype-owner-microsoft-twe.html new file mode 100644 index 000000000..bc7a04a36 --- /dev/null +++ b/_posts/2011-05-10-top-skype-owner-microsoft-twe.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Top Skype.owner = Microsoft tweets. +published: true +date: 2011-05-10 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/5361560330/top-skype-owner-microsoft-tweets +posterous_slug: post/5361560330/top-skype-owner-microsoft-tweets +--- + diff --git a/_posts/2011-05-13-testing-sux.html b/_posts/2011-05-13-testing-sux.html new file mode 100644 index 000000000..39606f678 --- /dev/null +++ b/_posts/2011-05-13-testing-sux.html @@ -0,0 +1,31 @@ +--- +layout: post +title: Testing sux. +published: true +date: 2011-05-13 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/5442267731/testing-sux +posterous_slug: post/5442267731/testing-sux +--- +

Maybe its my lack of usage, maybe its my lack of desire, but I hate writing tests.

+ +

Its not because I’m not good at it or that its hard to do, but it sucks, pure and simple. Why?

+ +
    +
  1. +It kills motivation to start something. It seems a massive undertaking, especially at the outset of a project to setup a decent testing structure for projects (somebody please make this easier). If it didn’t take knowing how to set up all my frameworks, gems, helpers, factories (or spending hours looking up docs about it), maybe I’d be more inclined, but its just daunting.
    2. +
  2. +It eats up dev time. I like programming, playing with code, tinkering with something until it works. Once it works, I can refactor. I could care less about tests. Granted, it ensures going forward my code works, but I know it works (for now). I tend to to CDT (code-driven testing) because of this.
    3. +
  3. +Often times it requires me to write more code than I did for the initial feature. Do I even need to say why I feel like this is not cool? Why would I want to spend more time writing tests than writing fun code? This code will never even see the light of day!
    4. +
  4. +Running tests eats time. It has been called the compiling of interpreted languages. I hate sitting and waiting for things to run. I will admit, CI helps with this. I like submitting tests, even if they break, to CI so it can run and I can write code.
    5. +
  5. +Frameworks require learning time. There are too many libs with too much complication out there. This contributes to #1 and the difficulty in learning, reading, understanding new code and the test framework that was used to test it. Jasmine for testing Javascript code is an incredibly simple library has been really easy to learn. All testing frameworks should take heed.
    6. +
  6. +Testing logic can get complex. Writing tests to test complex logic can be, well, complex.
    7. +
  7. +Tests require you to write testable code. Sometimes elegant solutions are not easily testable. Depends on how you test (blackbox, whitebox…).
    8. +

All that said (and I may have more), I test. With the best. Consistency over convenience I guess… It does ensure that your code will continue to work as expected. And that is a valuable thing.

+ +

It still sucks.

diff --git a/_posts/2011-06-02-better-testing-framework-is-b.html b/_posts/2011-06-02-better-testing-framework-is-b.html new file mode 100644 index 000000000..08e2dac86 --- /dev/null +++ b/_posts/2011-06-02-better-testing-framework-is-b.html @@ -0,0 +1,16 @@ +--- +layout: post +title: Better testing (framework) is better +published: true +date: 2011-06-02 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/6108505609/better-testing-framework-is-better +posterous_slug: post/6108505609/better-testing-framework-is-better +--- +

Not long after my last post, our group made a move to use RSpec more than we previously had. We have written (and in some cases still write) a LOT of our tests in Cucumber, which is a plain-text, behavior-driven syntax. At the same point that we made this change, I had the chance to update some code on another couple of projects (that we were using) and needed to update the specs. RSpec specs. It was really fast and in about 8 lines of code, I had them done.

+ +

Because of this, I realized something: RSpec is really simple! Cucumber tests require me to write big complex setup steps, helper steps, parser steps… steps, Steps, STEPS! Rspec? Not so much. There is some initial work but in the end, it was a whole heck of a lot easier to write in a Ruby way, rather than an English sort of way. This is always one of the biggest debates back and forth in the Cuke/no Cuke worlds.

+ +

From a developer standpoint, I want to write code. I want to write in the language I’m used to writing with every day. Plain-text is not as intuitive, because you have to make that plain-text mean something. It’s like writing your tests twice. They say that it’s great if you’re running the specs by non-technical people, but in our case and, I’d guess in 90% of the cases out there, we’re not. We’re just doing it because it seems easier initially.

+ +

That’s really all I had on that for now. Feedback on this one is definitely welcome.

diff --git a/_posts/2011-06-14-bettering-rails-pagination.html b/_posts/2011-06-14-bettering-rails-pagination.html new file mode 100644 index 000000000..f2a78a77f --- /dev/null +++ b/_posts/2011-06-14-bettering-rails-pagination.html @@ -0,0 +1,16 @@ +--- +layout: post +title: Bettering Rails Pagination +published: true +date: 2011-06-14 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/6528715761/bettering-rails-pagination +posterous_slug: post/6528715761/bettering-rails-pagination +--- +

Got assigned an interesting problem today involving pagination with nested routes. We do a lot of nested routes, for example

+ +

When you go to paginate, the initial page url looks right (e.g. example.com/companies/my-company/activities). However, the pagination links (for both WillPaginate and Kaminari) end up formatting so that we end up being sent to /activities?company_id=my-company&page=2, which works, but is not very consistent or user-friendly. What to do?

+ +

I researched a bit, and found the following StackOverflow question/answer,

+ +

This ought to be a default deal in any pagination solution out there, but it does not appear to be. It’s just nice that the paginators are so flexible. It would be nicer if they worked by default though… Fork time.

diff --git a/_posts/2011-06-20-listening-last-week.html b/_posts/2011-06-20-listening-last-week.html new file mode 100644 index 000000000..f6d86c8e8 --- /dev/null +++ b/_posts/2011-06-20-listening-last-week.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2011-06-20 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/6724483902/listening-last-week +posterous_slug: post/6724483902/listening-last-week +--- +

[[posterous-content:hmnyahuwdIEbetrrwCbb]]

Listening last week

diff --git a/_posts/2011-07-13-mostly-false-ive-been-listeni.html b/_posts/2011-07-13-mostly-false-ive-been-listeni.html new file mode 100644 index 000000000..ffdf0f4b8 --- /dev/null +++ b/_posts/2011-07-13-mostly-false-ive-been-listeni.html @@ -0,0 +1,10 @@ +--- +layout: post +title: Untitled +published: true +date: 2011-07-13 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/7559745380/mostly-false-ive-been-listening-to-this-for-the +posterous_slug: post/7559745380/mostly-false-ive-been-listening-to-this-for-the +--- +

[[posterous-content:CwEHJAGkIeadjigxjChh]]

Mostly false. I’ve been listening to this for the past three days straight. Radiohead is soooo good.

diff --git a/_posts/2011-07-13-why-osx-lion-is-a-consumer-dr.html b/_posts/2011-07-13-why-osx-lion-is-a-consumer-dr.html new file mode 100644 index 000000000..ea994c801 --- /dev/null +++ b/_posts/2011-07-13-why-osx-lion-is-a-consumer-dr.html @@ -0,0 +1,18 @@ +--- +layout: post +title: Why OSX Lion is a Consumer-Driven Update +published: true +date: 2011-07-13 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/7559599202/why-osx-lion-is-a-consumer-driven-update +posterous_slug: post/7559599202/why-osx-lion-is-a-consumer-driven-update +--- +

I started using the developer release of Lion about two weeks ago. While I enjoy a few of the new features, I have not found it to be as impressive as Snow Leopard was in it’s offerings. This may seem odd because Snow Leopard was branded as a “maintenance upgrade” focused on stability and performance improvements. Well, that update that we were told was “maintenance” was a massive architecture change. OS X went from simply supporting 64-bit to actually running the entire OS as 64-bit. It added better multi-processor/core support and GPU usage improvements. Those changed the way any and all of my software runs.

+ +

Not being a Mac Dev, some of the Lion additions don’t mean much for me and the software I write and run or the software that I use on a daily basis. Moreover some of the changes (fullscreen esp.) don’t help me in the least when I’m running multiple monitors. The idea sounds nice, but in reality, when am I ever using one window at a time? Even as I’m writing this I’m checking Twitter and Propane (Campfire) for other info.

+ +

Some of the changes are useful (though have yet to be integrated into my workflow) like Mission Control. I struggle to regularly and effectively use multiple desktops, especially when I’m again using multiple monitors. It’s just not as time saving for me. The fact that Apple finally seems to have gotten Cmd-Tab to work right over multiple spaces will probably change my mind on this fact.

+ +

All this goes to say that most of these changes are UI updates. They are very driven towards making the UI look closer to the iOS UI. Not a full conversion, but Apple repeatedly states that it is the “full circle”, bringing back to the Mac what it did to the mobile UI space. Which is not a bad idea. However, it does make it more about what the user sees than how it performs, and that makes it very consumer-driven. Lion makes Apple’s OS lines (mobile and desktop) more unified, easier to transition between and, most importantly, more attractive. Users, technical and non-technical alike will find these changes revolutionary and useful. While these are all things I appreciate and will use often, as a non-platform specific developer, not as useful to me in every-day application and/or development.

+ +

Snow Leopard will probably stand alone (or at least very lonely) as a major update that did not involve a massive UI overhaul and yet revolutionized the platform. Apple still has a massive leg up on most other platforms I have used (Windows, and (oh snap) Linux) simply because they are dedicated to improving UI and the OS performance at the same time. I have not used a product so well polished in both areas.

diff --git a/_posts/2011-07-14-first-jquery-hack.html b/_posts/2011-07-14-first-jquery-hack.html new file mode 100644 index 000000000..de87d8218 --- /dev/null +++ b/_posts/2011-07-14-first-jquery-hack.html @@ -0,0 +1,79 @@ +--- +layout: post +title: First jQuery Hack +published: true +date: 2011-07-14 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/7629350472/first-jquery-hack +posterous_slug: post/7629350472/first-jquery-hack +--- +

In the last 2-3 days I’ve hacked a couple of additions onto two pretty important JS libs for Rails: rails.js (not super substantial) and jQuery. It was a lot of fun to learn the guts of these widely used and not often tampered with libs. As I’ve been doing so much js scripting of late, it’s gotten really easy to read through js code and understand functionality quickly which is a real help. And when it comes time to patch, finding where it needs to go is simply working through the code flow. The more I contribute to OSS, I find that the best code documentation is the code itself. There’s almost nothing so good as good code.

+ +

I’ve never really delved into the heart of jQuery before, especially over something so central as $.ajax. I’ve done the occasional lookup when some of my code is acting up, but today was the day I needed functionality that didn’t exist.

+ +

As referenced often before in previous blog posts, SalesCrunch is using a wide diversity of technologies to move people around, especially during a live meeting. It’s heavily centered around Node.js and Redis to handle async messaging between server and client. Because of this, there are a lot of failure points for us, and without having access to client’s browser info, it can be hard to debug beyond the stack traces we get. So I was tasked with setting up a diagnostics page, where you could check all the variables and connections that our servers make to your browser.

+ +

It looks something like this:

+ +

[[posterous-content:qGwIbbbczJatqpsGjdId]]

+ +

We cover a lot of the items we require to use our software: JS, Java, Flash, Socket connectivity, and so on. It was fairly easy to tackled most of them.

+ +

Flash: Simply checked if there is a flash plugin installed via the browser’s navigator.plugin object. JS is simply populating the field with enabled… via javascript. If it’s disabled, it uses <noscript> tags to populate. Java is a mini-applet that simply provides information on the JVM. Browser and OS info can be determined from the HTTP user-agent request header. We can query our socket (Socket.IO) to see if it’s connected or not.

+ +

The last one, bandwidth test, was the most interesting, mostly because it involves timing and math skills, both of which I generally fail at. So I inquired of google for some ideas. The general idea is fairly simple: Download something of a known file size and time it. Bandwidth, is after all Mbits per second (not to be confused, as I did earlier with MBytes… ehem). This should be easy then.

+ +

The trick is: It’s not entirely easy. Everything you hear about trying to calculate bandwidth by this method is working around server response delays. You can see it pretty quick in Chrome when running locally:

+ +

[[posterous-content:voBrdwHqwznsJsAzDwzs]]

+ +

Now, the upside is that you don’t have to worry about this as much, so long as your file is big enough to make delay negligible and not count for much against the total download time, like so:

+ +

[[posterous-content:HvfdsGjdniDfvddthslt]]

+ +

As you can see, the waiting time is not a big percentage of the overall download time. However, I wanted to do better, partly out of curiosity, partly out of pursuing excellence. Since I’m using jQuery, I figured it shouldn’t be too hard to track down a way to detect the first response back from the server. More importantly, I want to know when the actual file data (in this case, a screenshot from Minecraft) is being returned, minus the HTTP header info. Precious bytes…

+ +

Digging into the jQuery code, they do, in fact track a readyState, very similar to the one specified by XMLHttpRequest. This has a number of important states that we care about:

+ +

(from http://www.w3.org/TR/XMLHttpRequest/#states

+ +
+

The XMLHttpRequest object can be in several states. The readyState attribute must return the current state, which must be one of the following values:

+ +

UNSENT (numeric value 0) + The object has been constructed.

+ +

OPENED (numeric value 1) + The open() method has been successfully invoked. During this state request headers > can be set using setRequestHeader() and the request can be made using the send() method.

+ +

HEADERS_RECEIVED (numeric value 2) + All redirects (if any) have been followed and all HTTP headers of the final response have been received. Several response members of the object are now available.

+ +

LOADING (numeric value 3) + The response entity body is being received.

+ +

DONE (numeric value 4) + The data transfer has been completed or something went wrong during the transfer (e.g. infinite redirects).

+
+ +

Now XMLHttpRequest spec calls for an event to fire when these states changes. It’s called onreadystatechange. Simple enough. Except for the fact that jQuery ignores and stuffs this event into a sock and leaves it for the moths in a dark closet somewhere, meaning simply: It kills this event and doesn’t fire one of its own, stating:

+ +
+

No onreadystatechange mechanism is provided, however, since success, error, complete and statusCode cover all conceivable requirements.

+
+ +

The hell it does.

+ +

The problem with jQuery’s “all conceivable requirements” is that my requirement is not “before I make the request” or “after the request is complete”. It’s somewhere in the middle after I’ve made the request when I’ve started to receive data back from the server. To be specific, this is state 3: LOADING. So how do we capture this info?

+ +

Its actually pretty simple. jQuery, despite not sending an onreadystatechange event, still tracks the right states. While I probably ought to spend time writing the correct event, responding to the states is a bit simpler. So on the current jQuery codebase (on github, I added the following optional callbacks (github commit).

+ +

So if the states are 2, call receivedHeaders, if 3, call loading. These methods can be set on the $.ajax method like so.

+ +

This way we can then grab the startTime for the download when we are receiving the known data back from the server (loading) and the end time on success. In the end, this was more than we really wanted to do, since it is unlikely that jQuery will accept this patch into the code base as it explicitly claims no support for onreadystatechange. Therefore, we’re using vanilla jQuery and allowing for error in our results. It’s not like it needed to be perfect. I just ended up being curious.

+ +

One other thing to note (pointed out to me by @tjsingleton) if you are doing what you ought to do and make multiple requests so that you can average the results, you will need to stagger your async requests, otherwise your browser may get tied up trying to make too many requests at once and add more artificial delay. In other words, don’t request all 10-50 samples at once. I staggered the requests to two at a time for ten requests total for a decent average. I used jQuery’s deferreds to trigger more requests once the two images have completed. This wouldn’t be necessary, however in the case that you used the above method to capture start/end times once actual data is being retrieved.

+ +

I may generalize the code and post it at some point in the future since it’s pretty nifty.

+ +

Edit: I’ve submitted my initial patch to jQuery for consideration. You can track it here and on the jQuery bug tracker

diff --git a/_posts/2011-08-18-designing-for-change.html b/_posts/2011-08-18-designing-for-change.html new file mode 100644 index 000000000..945c8ae04 --- /dev/null +++ b/_posts/2011-08-18-designing-for-change.html @@ -0,0 +1,44 @@ +--- +layout: post +title: Designing for Change +published: true +date: 2011-08-18 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/post/9067560580/designing-for-change +posterous_slug: post/9067560580/designing-for-change +--- +

I promised some people via twitter that I was going to post again on discoveries regarding client-side rendering, but this is not that post. Instead, I’ve decided to branch into a mix of design and engineering related thoughts that I’ve been working on as a result of some interesting issues while going through the latest design iteration of several of the interfaces of SalesCrunch’s CrunchConnect platform.

+ +

A Real Life Example

+ +

Our product, as I have noted before, is a presentation and tracking tool, purposed to allow users to present slide decks, normally as part of a sales presentation and provide the presenter with tracking metrics both in live (presenter-driven) and emailed (viewer-driven) meetings or presentations. As part of this, we allow the users to upload and/or select decks for use in presentations or to send to users. We have an interface for doing this that used to exist in two parts: One for uploading, naming and organizing decks, the other for selecting decks for use. As part of one level of redesign, these two interfaces were forced together so that, during a live meeting, a presenter could both upload and then use a deck during a live meeting. The result looked something like the following two screens.

+ +

[[posterous-content:GdzJgHItbfFEhCniAsbq]] +Image 1: The Original Selection Screen

+ +

(Please ignore the design flaws, some of these shots were of prototype interfaces.)

+ +

The main thing I want to note here is the input and submit controls. In the above shot, we have two main controls for selecting a deck: A <select> element for selecting a deck and a select button for submitting the choice. These elements are within an ajax-form that submits the chosen deck for use in the existing context. Not too complicated. The biggest challenge of this interface is that we need to know the deck id, title and unique slug. These three pieces are needed for either UI updates after the deck is selected or to send back up to the server to track what deck is being used in the presentation. In the select, the id and title are in the standard <option> value and text values. The slug is stored as an html data-* attribute on the option for easy retrieval later. So this is fairly straightforward, and the systems surrounding the UI updates and server actions are pretty simple.

+ +

The problem shows up when our design/product team decides the interface is not very straightforward, and I agree, mainly in the separation of the select/upload flow. Our new design separates out the two into separate tabs as seen here:

+ +

[[posterous-content:yquovwDcJeokGfyiJhGF]] +Image 2: The New Selection Screen

+ +

This makes it easy to see here are my decks and if I need more, I go to the upload tab, there are no major redirections or interface changes.

+ +

The problem is the redesign moves all of the data we depend on to update the interface around. Note the input and submit count change. We go to zero inputs and n submit buttons where n equals the number of decks. Moreover, we actually lose the form altogether and go with ajax-actioned buttons which send the same action as the form in the end. So we’ve just added a much larger number of submits/actions for the user to choose from. While I could debate the value of this idea, the point of this post is elsewhere.

+ +

This new design has not just changed the page layout, but also the landscape of where the information is stored on the page. Now instead of being in a single input, the data is scattered to the individual submits. This means we have to change where we go to retrieve all the information needed to update the ui and inform the server. That means hours of refactoring JS UI handlers/submitters on top of the extensive html UI changes.

+ +

So my thought is this: How can we design a screen like this to make the UI wildly flexible while ensuring the backend will work the same (or at least nearly the same). UI tied up with large amounts of data means double the work untying it from the data so that we can tie the data back to new UI elements. Part of the answer lies in the classic separation of UI and data in the M and V pieces of the MVC design pattern. Embedding a lot of data within a view is already a bad idea. It’s partly unavoidable for some client-side purposes, but there is still a need for keeping process-dependent data out of specific view elements. Perhaps if we just used ids within the view and link back to the data store within JS objects. Backbone.js is an excellent library to aid in keeping this design paradigm for JS-heavy interfaces.

+ +

The other piece of the answer lies in simply designing for change. Designers love to tweak things, move elements around, change entire screens for the next design iteration. In the case of this design change, our inputs and submits change a great deal, but the server-provided data and the data expected on submit is exactly the same in both cases: id, title and slug. Only the layout changed. If we design for change, all we’d need to do is move the html around, fix up some css and hopefully tie a minimal amount of data to each new element so that when the submit is clicked, the same data is collected for the UI and server updates as when we had a single input/submit pair.

+ +

What are some strategies for doing this?

+ +
    +
  1. As mentioned, store the data off of the view, in a “persistent” JS collection of some sort. This could be retrieved and stored along with the initial server page render or in later ajax requests as the data is needed. In our case, we collect most of the deck data in an “infinite scroll” method, only generating deck rows as you scroll down a few rows.
    2. +
  2. Let the server handle the data and just request new info as needed. No need to store anything client-side if you don’t want, just submit an id and let the server figure out what happens and have it render and push back up the update UI. I don’t recommend this as much as it increases server traffic and most of this stuff can be handled with data retrieved on the first page request or subsequent “page initialization” requests.
    3. +
  3. Store the data on the page and pull it out as needed, just don’t tie it to a single element. HTML5’s addition of data-* attributes is an awesome way to store stuff on page, but when you tie it to changing elements, it can be a nightmare to move things around. Instead, put the data somewhere on the page, the body or other non-visible elements for later retrieval so that if the page changes, the data will still be available.
    4. +

So the next time your product/design team says “Can you radically alter the design for this view?” you should be able to tell them “Give me an hour”. When they ask you how you can do that in such short order, you’ll be able to tell them it’s because you designed it to change.

diff --git a/_posts/2011-09-13-ie-gotchas-ie7-and-z-index.html b/_posts/2011-09-13-ie-gotchas-ie7-and-z-index.html new file mode 100644 index 000000000..b72963dc9 --- /dev/null +++ b/_posts/2011-09-13-ie-gotchas-ie7-and-z-index.html @@ -0,0 +1,30 @@ +--- +layout: post +title: ! 'IE Gotchas : IE7 and z-index' +published: true +date: 2011-09-13 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/ie-gotchas-ie7-and-z-index +posterous_slug: ie-gotchas-ie7-and-z-index +--- +

We have a lot of layers going on in our meeting UI. Fortunately, most of it is not overlapping. But in the odd case where we need to overlay something and it’s not taking over the whole page (like a lightbox or other popup), we have to ensure that it layers properly. Usually this is done with a mix of position: absolute and z-index. However, this is not the case in IE7. IE7 has this funny issue where it starts a new zero-indexed layer with each positioned element (detailed at quirksmode.org). So you do this:

+ +
+ 
div.parent { position: relative; }
+div.parent div.child { position: absolute; z-index: 100; }
+
+ + +

and the layer as a whole, is still at z-index 0! So everything within the parent element will still be below other page elements

+ +

Oh snap.

+ +

The fix is simple. Give the parent element an even higher z-index. So it becomes:

+ +
+ 
div.parent { position: relative; z-index: 110; }
+div.parent div.child { position: absolute; z-index: 100; }
+
+ + +

This should give you the proper layering.

diff --git a/_posts/2011-10-05-ruby-and-rails-described-perfectly.html b/_posts/2011-10-05-ruby-and-rails-described-perfectly.html new file mode 100644 index 000000000..d185da76b --- /dev/null +++ b/_posts/2011-10-05-ruby-and-rails-described-perfectly.html @@ -0,0 +1,15 @@ +--- +layout: post +title: ! 'Ruby and Rails Described Perfectly:' +published: true +date: 2011-10-05 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/ruby-and-rails-described-perfectly +posterous_slug: ruby-and-rails-described-perfectly +--- +
+

The problem with object-oriented languages is they’ve got all this implicit environment that they carry around with them. You wanted a banana but what you got was a gorilla holding the banana and the entire jungle.

+
+

seen on SaturnFlyer

+ +

from: Coders at Work

diff --git a/_posts/2011-10-26-i-love-specs.html b/_posts/2011-10-26-i-love-specs.html new file mode 100644 index 000000000..9916d58bd --- /dev/null +++ b/_posts/2011-10-26-i-love-specs.html @@ -0,0 +1,16 @@ +--- +layout: post +title: I Love Specs +published: true +date: 2011-10-26 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/i-love-specs +posterous_slug: i-love-specs +--- +

Alternately titled: How I learned to hate the player, not the game.

+

 

+

So I was wrong. There, I said it. Specs are awesome. When written write, in a good language (because fuck Cucumber man), and in isolation, they are a beautiful thing. I've really started to dig TDD. What changed my mind? 

+

Well a few things. One to many changes implemented with unexpected consequences to the rest of the system was a big one. Which, that in itself is another problem altogether. That is a problem of badly coupled code, a problem we are working to resolve elsewhere.

+

Another is being able to better flush out the spec. Our specs are written by technical people, but often times, the implementation details don't show up until you're, well implementing the damn thing. But this comes a step earlier if you write tests first. Tests tell you how it ought to work, and when you realize you don't know how it ought to work, you have to ask. So you clear up a lot of misunderstandings before you are in the middle of writing the feaure and realize it all has to be redone because your understanding of the implementation was wrong.

+

Lastly, it gives you a reference point to say "Here's how it worked, was supposed to work, or how it will work." Then you can go forth and change either the spec, or the code (one at a time please), and ensure that it either continues to work as before, or that the spec has changed and needs to now represent the new functionality.

+

In any case, I am a reformed developer, now a believer in testing and TDD. Go forth and do likewise.

diff --git a/_posts/2012-02-26-choosing-a-path-is-hard.html b/_posts/2012-02-26-choosing-a-path-is-hard.html new file mode 100644 index 000000000..a749eaf3d --- /dev/null +++ b/_posts/2012-02-26-choosing-a-path-is-hard.html @@ -0,0 +1,41 @@ +--- +layout: post +title: Choosing a Path is Hard +published: true +date: 2012-02-26 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/choosing-a-path-is-hard +posterous_slug: choosing-a-path-is-hard +--- +

Today I read a number of articles in succession, as I often do when reading on the internet. Today's internet articles are so scattered with hyperlinks and youtube videos, I feel like someone loaded up a shotgun with other content and blasted it into what I was trying to read.

+

In any case, these three articles are posted here. Try to read them first, even if the content isn't really along your lines of work or interest. In order of my reading (and linkage):

+

Giles Bowkett summons monsters

+

which references:

+

Rails Went Off The Rails

+

which quotes a bit from:

+

A Conversation with Alan Kay

+

Okay, so that was a lot of reading. Each one sparks something a little different in me. The first article focuses the second, in that it talks about what Giles is getting at in code and programming culture. His assesment is fair in that:

+
+

Once you view the post as being about software development culture instead of about software development tools, everything clicks together. This is not a post about Rails and Node, it’s a post about people and culture. It’s not about how Node and CoffeeScript change everything, it’s about how they change nothing, because we haven’t changed. We’re still on a path to re-invent all the ceremony and design patterns and cruft of the last go-around.

+
+

That is depressing to me, partly because it's so unbeliveably true. I am just as guily of what Giles talks about in his article, that the old decry's the new as bratty and immature while the new mocks the old as slow and inefficient. To quote Giles: 

+
+

The sulky, bratty teenage Rails had three big things it made fun of the Java old folks for: design patterns, the JVM's long startup time, and overly ceremonial code which had nothing to do with the day-to-day experience of writing web apps. Today Rails has a startup time which is so long, the Rails community actually created Spork, a gem which keeps your Rails environment alive between test runs, because otherwise it's hard to use TDD with Rails. Rails tests can take so long to run that by the time a test run completes, you're liable to forget what you were doing, and, in fact, your name. The Rails community mocked design patterns relentlessly when it was Node's age, but look at how ActiveRecord::Base hijacks inherited and you'll realize that this [...] is an ActiveRecordClassFactory in denial.

+
+

Both of these views make me sad, for a few reasons. One is that it shows how true the first quote I posted. We are reinventing things because programming is becoming a pop-culture career that likes to cleave to the new, the hip, the easy. Secondly, because of this, we're redoing a LOT of things that are already done, designed, patterned really well. (Even worse, we're doing it in JAVASCRIPT!?) 

+

And then you read an article with men like Alan Kay and Stuart Feldman who embody what it meant to be a computer scientist, not just a developer. In a sense, they understood keenly the "scientist" in the Computer Science field. Just look at what they did: They spent the majority of their careers in research departments, not developing software as a product, but inventing languages, creating methodologies, documenting patterns.

+

And somewhere in a world gone crazy with the latest tech, the hottest new language on the block, which should lead to the sexiest job in the world, we've lost the drive to create, rather than be a creature governed by our tastes. As the oft-quoted and still missing _why put it: 

+
+

when you don’t create things, you become defined by your tastes rather than ability. your tastes only narrow & exclude people. so create.

+
+

This "being defined by tastes" attitude is abundantly clear in the Rails and Node communities, as evidenced even in the flamewars had around such blog posts. "We do things our way because it's better because of X, Y and Z and you suck at all those things" rather than spending that energy in creating new. I've jumped on that bandwagon on several occasions. And I am not proud of that. And sadly, as I read articles like these, I realize that in most of my day-to-day development, I am not an engineer or a scientist. I am a subscriber to pop-culture, merely a user of what exists, defining my work by what I use. Alan Kay puts it eloquently:

+
+

If you look at software today, through the lens of the history of engineering, it’s certainly engineering of a sort—but it’s the kind of engineering that people without the concept of the arch did. Most software today is very much like an Egyptian pyramid with millions of bricks piled on top of each other, with no structural integrity, but just done by brute force and thousands of slaves.

+
+

This upsets me, because it means I am not advancing the science of computers, but rather abetting the increase of knowledge and craft. I am a specialist, not a scientist. And as we all know from DJ Horace "the Pacifist" Walsh from K-JAH radio in GTA3, "You want knowledge, get yourself the Scientist", mon.

+

What is interesting to me in this pop-culture as it moves from popularity towards acceptance is the reverence of old languages and design patterns. Ruby, as it has matured (we're going on 16 years since initial release), has started to look back. Smalltalk, oddly, is a language rubyists hold up as an icon of the programming language world. Which begs the question first, why are you using Ruby then, and second, why now? Smalltalk has existed since the '70s and you're only now discovering it had good ideas? At the same time, patterns are again all the rage, which is what the Ruby/Rails community initially flocked to Ruby to escape from in Java. And the cycle begins again -- after it's barely started in Ruby -- with Node. From nodejs.org's homepage:

+
+

Node.js is a platform built on Chrome's JavaScript runtime for easily building fast, scalable network applications. Node.js uses an event-driven, non-blocking I/O model that makes it lightweight and efficient, perfect for data-intensive real-time applications that run across distributed devices.

+
+

There have been dozens of blog posts, articles and examples written about this very statement and how Node misses the point of what it is trying to deliver. But more important to me is the step back this framework seems to take in recreating concepts that have existed for years. You know what else accomplishes the same ideas? Threads. But oh wait, threads are hard. So here we are again, reinventing the wheel in another color to excite another generation of "ninjas and rockstars" when we could have been creating new ways of computing. It seems such a waste.

+

All this is to say, instead of getting depressed, I want to find a way to step off the circle and move forward, build something new, or at least learn from what exists and build from that. Aperture Science founder Cave Johnson said: "They say great science is built on the shoulders of giants - not here. At Aperture we do all our science from scratch; no hand holding." Let's not do that. Let's learn and move forward. Create, and build bridges. For science.

diff --git a/_posts/2012-02-27-bjarne-stroustrup-c11-style.html b/_posts/2012-02-27-bjarne-stroustrup-c11-style.html new file mode 100644 index 000000000..7561e7268 --- /dev/null +++ b/_posts/2012-02-27-bjarne-stroustrup-c11-style.html @@ -0,0 +1,17 @@ +--- +layout: post +title: ! 'Bjarne Stroustrup: C++11 Style' +published: true +date: 2012-02-27 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/bjarne-stroustrup-c11-style +posterous_slug: bjarne-stroustrup-c11-style +--- +

Worth a watch.

+

+ + + + + +

diff --git a/_posts/2012-05-03-very-helpful.html b/_posts/2012-05-03-very-helpful.html new file mode 100644 index 000000000..0c9690ac3 --- /dev/null +++ b/_posts/2012-05-03-very-helpful.html @@ -0,0 +1,11 @@ +--- +layout: post +title: Very Helpful +published: true +date: 2012-05-03 +categories: [] +posterous_url: http://blog.hungerandthirstdesign.com/very-helpful +posterous_slug: very-helpful +--- +

From the effervescent @jeg2

+

http://speakerdeck.com/u/jeg2/p/10-things-you-didnt-know-rails-could-do

diff --git a/_posts/2012-12-12-left.markdown b/_posts/2012-12-12-left.markdown deleted file mode 100644 index c909636ca..000000000 --- a/_posts/2012-12-12-left.markdown +++ /dev/null @@ -1,27 +0,0 @@ ---- -layout: post -title: Left -category: posts ---- - -Left is a clean, whitespace-happy layout for [Jekyll][jekyll]. - -This is designed to be an easy layout to modify for your own blog. It was -extracted from [zachholman.com][zh], which means it was battle-hardened from -years of posting serious blog posts about emoji and swear words. - -You're actually reading an example post right now. See how cool it is? Oh, be -quiet, you're making me blush. - ---- - -You'll want to [get the code][left] and read the README to learn how to -install and set up Left for your own purposes. - -If you have any questions, follow and ping me on Twitter- I'm -[@holman][twitter]. - -[jekyll]: https://github.com/mojombo/jekyll -[zh]: http://zachholman.com -[left]: https://github.com/holman/left#readme -[twitter]: https://twitter.com/holman \ No newline at end of file diff --git a/atom.xml b/atom.xml index 36f0b11f7..066f88ca5 100644 --- a/atom.xml +++ b/atom.xml @@ -4,24 +4,24 @@ layout: nil - - Zach Holman - - + + Luke van der Hoeven + + {{ site.time | date_to_xmlschema }} - http://zachholman.com + http://plukevdh.github.com - Zach Holman + Luke van der Hoeven - + {% for post in site.posts %} {{ post.title }} - + {{ post.date | date_to_xmlschema }} - http://zachholman.com{{ post.id }} + http://plukevdh.github.com{{ post.id }} {{ post.content | xml_escape }} {% endfor %} - + \ No newline at end of file diff --git a/posterous.rb b/posterous.rb new file mode 100644 index 000000000..7f765ccbc --- /dev/null +++ b/posterous.rb @@ -0,0 +1,85 @@ +require 'rubygems' +require 'jekyll' +require 'fileutils' +require 'net/http' +require 'uri' +require "json" + +# ruby -r './lib/jekyll/migrators/posterous.rb' -e 'Jekyll::Posterous.process(email, pass, api_key, blog)' + +module Jekyll + module Posterous + def self.fetch(uri_str, limit = 10) + # You should choose better exception. + raise ArgumentError, 'Stuck in a redirect loop. Please double check your email and password' if limit == 0 + + response = nil + Net::HTTP.start('posterous.com') do |http| + req = Net::HTTP::Get.new(uri_str) + req.basic_auth @email, @pass + response = http.request(req) + end + + case response + when Net::HTTPSuccess then response + when Net::HTTPRedirection then fetch(response['location'], limit - 1) + when Net::HTTPForbidden then + retry_after = response.to_hash['retry-after'][0] + puts "We have been told to try again after #{retry_after} seconds" + sleep(retry_after.to_i + 1) + fetch(uri_str, limit - 1) + else response.error! + end + end + + def self.process(email, pass, api_token, blog = 'primary', tags_key = 'categories') + @email, @pass, @api_token = email, pass, api_token + FileUtils.mkdir_p "_posts" + + posts = JSON.parse(self.fetch("/api/2/sites/#{blog}/posts?api_token=#{@api_token}").body) + page = 1 + + while posts.any? + posts.each do |post| + title = post["title"] + # slug = title.gsub(/[^[:alnum:]]+/, '-').gsub(/^-+|-+$/, '').downcase + posterous_slug = post["slug"] + slug = posterous_slug[0..44] + date = Date.parse(post["display_date"]) + content = post["body_html"] + published = !post["is_private"] + name = "%02d-%02d-%02d-%s.html" % [date.year, date.month, date.day, slug] + #name = "%s.html" % [slug] + + tags = [] + post["tags"].each do |tag| + tags.push(tag["name"]) + end + + # Get the relevant fields as a hash, delete empty fields and convert + # to YAML for the header + data = { + 'layout' => 'post', + 'title' => title.to_s, + 'published' => published, + 'date' => date, + tags_key => tags, + 'posterous_url' => post["full_url"], + 'posterous_slug' => posterous_slug + }.delete_if { |k,v| v.nil? || v == ''}.to_yaml + + # Write out the data and content to file + File.open("_posts/#{name}", "w") do |f| + puts name + f.puts data + f.puts "---" + f.puts content + end + end + + page += 1 + posts = JSON.parse(self.fetch("/api/2/sites/#{blog}/posts?api_token=#{@api_token}&page=#{page}").body) + end + end + end +end \ No newline at end of file From ae2446a6355543f2ddb715dfce74c7e14ac2a873 Mon Sep 17 00:00:00 2001 From: Luke van der Hoeven Date: Tue, 18 Dec 2012 07:49:04 -0500 Subject: [PATCH 3/6] update fav --- apple-touch-icon.png | Bin 43521 -> 0 bytes favicon.ico | Bin 1150 -> 1406 bytes 2 files changed, 0 insertions(+), 0 deletions(-) delete mode 100644 apple-touch-icon.png diff --git a/apple-touch-icon.png b/apple-touch-icon.png deleted file mode 100644 index 052bb988369d627b1a65411b50f2253684bfebb7..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 43521 zcmV({K+?a7P)4Tx0C)j~RL^S@K@|QrZmG~B2wH0nvUrdpNm;9CMbtL^5n^i$+aIn^?(HA4aZWV5ov6ELTdbo0FI&wK{O>*+w4vx20?>!`FrQsdJlnHR>OPy zcd~b_n$otK2Za4V;76L-DzNVtaSB-y0*E}{p()372;bw_^6ZZ}PI-92wGS&j#91PI zKs7DSe@(bk%_Y-7gGe}(^>I=@oY#w#*Bu9GZf3^F5WP>3rn}7Ut74&?PWBFvy`A)a zPP5)V!Xd&78LdA?xQ(9mjMYElVd13a#D+Z_7&Y|xU=_C-srWU*6kiZcC!$nw*)9$7 zn6CX+@=AhmkT}X@VSsa5NKe;HZuq)~1$`#h6R+ZTR#D-3j}vF!)ZOnz+5)dI4jl{{ z44Mr{P!L4~VVJN`K!!XTF*LGrKO?IK8z<8w`3e3jI8lUGNUta*C8 zn(P`s>{pjD=7Kek#B;Fw@hxAK%$F&Q6vg9J^Xf~4by_hu-=A!MJ3Znq&n~srbFGPs zH&&aMXZ>nO`|hf|ljc?VPhR!${AbO?W8x_>CU%PFA&Hm8F7cAsOREdwU~R_;ot1_u z(ruCYB-LPGn!NQdT|ZlRy+(fw^-+`=%+gee_kY4FWHg<*4sZI8+sFJD270UUORdLHO0nA4V) z%{fwsET5CQ>B?eK%uw4yQc~9?*JVo2}ze(;aRcp*ceL#HUJSllrgm5wQKR zQu+C;QrUh^8rFfA`ftFz{YAidi-`aL010qNS#tmY3ljhU3ljkVnw%H_03ZNKL_t(| z0ldB0k7Y}ir)6jI{oOp}PSIcn9%_c;T->NWAcdgplfm z7eqH?b#`TD<#6xKJI8P%Zp?#+&*M4A`M#f>9TAz8BUQzYIDYoo+qT)VwU({Pmf3Fo z`G4>iXPr)`wA<}`N@?-kDy>etw9fd|;+zj9N{j2I(=DBLr+D~GhtG)fuH`9rHX4;q z%d%|?qzsoi_~YZ$+4cgTZnw*Q>*6BMd!Dv{r>V;(+il8G-uA3xbQZu$*Wa9<9-oxM z!^5((w_Bb(e^w^?{jza=qs-6Dl>Wp->GpcXKAusww%IO2;TYG~qNAR-fIIMT?@X9% zEPwem^7mQ#Wj*%uY&ZfIde5y9zf+vA;c$>^wkfPFG(X4NeDq4^r=NU>WLar$f-wK3ErU(;t7f zyU@1RDO)eL%4eT_T279SgNwPj>2mME!?L%3kUkG6-|kj?w`q5FX0}X;SMb*6`4M%> zFCOwvHw=(pfNy90h;X_S1_L8`*ef!DOv8CC6a4vDjvz~B5n$>Vc_W>XIW59BoFegF zKF@vI@m@WIu6Ek97RVRI%1rzgzEnnC{A<&R(whLWgVLj;6BE61401$pcw=0zjFGm5 zpU;@iY0^5!h4V%0^jW5m02N`iJmn-vKErek2pYc;kG=X%QoSZ8AhesK~5|Cm!Jnys|81=bvC)0kawx6_p#;~h6soyej zWeVW{wlkyvX~@5VakU-<8nTKa&laDc6PWgOaZGLADxup zsnQEcPd&EnA-t6_EBrhJb5%Nk*_81cL0CtH!!9>?vPv;W2^FFQ1-GfWGJ35*qKd2f z%bv==;?@qNjWGw)3Xp|h>0EI3@|tzFg>$TueAvgw5an0e;!2gMwek+k=%G08O>|55 zq!R{{z(+@iVMvi_ZMLlhQjFU_MpO`OO)dM(M^dy?zCw^8WCyw4NDY}h#HK$XgaG8A z1C)!oKpl27jN_O0sV6n%20tA+MP^fS`K(X-956^Zq)k<0wqtqjbKU0w(S z^MjgT8q0xSrAdoH&{mSFay4d7X+A~LfGh+5*<|-wCBNEv;tP8OubTxLWB87M|h?~jecw=cu+Eh(Skn8 zRJLfIu_o_uBj46pOV;ZGSg(0f&npa$27Zt0vyFyHE$`uXht6wDi3Fv?-rgRFIE{K# znsr;+bM`!yr;dsoFvK-$id0RF{t1>-w>Y$o*i&4ueXISpq|ZjRa25n3LM481FVpa> z&qN*D77s|kkyPUQ72c=^hRTC^rV|3o%xNS<=TxLK5Jm&8TiCL(=f={F`!ZG@0+tL4 z%0&n|FqCBJaBa#TC)j9F@w7iU#hx3S0yFn18XL+%8Mx~+_A~vk@8f~d(97T5hC{;u zW5%c*7Xh{ol29SCPzXrvks91%(Bw$5AktA(B|;d0yNw!2O(`gnYu(o8rvemg@moC6 zmhDK4)GHKRw`UZ11v6K<)ale~s$|vS)YMctz?he@8e&>km;2UTQPcFVKC6_qUm5At zr;X+_N<~#v1|X{G)ggRUJgC0xZwCgBs+4%Kiz+UBM$m^c4FFqXm-6I78N3^7hCqqmj1UiKNK8RrZ)Xk>o}sdWf#fQWnb`Bo1+`^a&H z*FIW*gJ1bi5oo`LCo&#&f*TKUg)HdoY*S(jvs<~n@=}3b)J41(w z>NE75=0&`8YCg4Tp34BT4(japd*u`yRT{jqRb`TWYlBH@UuD>F6KQZfAeBxny>99_ z9TN{Y@Yp4P^RQ2(^@=d1l(Ith_m2IcAyr^=O@I``x|kWkK~0LMvwDag)?_VVrVOXb zkhMeHp`NC5AxcWW>WI%as!F6eij{^90sNh?g3N_yD6A)pG8sjNd0Dqiz$lbS(s$t# z1so@MHFmAXzfM=%5-EYP94SYc7`+f*l^9sYF#nR1fw9=}excCql>h;SbShXR_ z+1GBj7inbU5>e@y@XojnZLm{~cK!usn8k2IWy4x0l~w$Llc9A0W2f5>^Z5JhjJ^$o z#ePsbbr{B(=uw7tgjbyO`x7t|aACL?Osm0&@Wly?PQO8T8WS_(<%~frU4+BF$z0lR zQH2#ON0$#td`ege9Fa>Y(2PC-_ByD6!F)!sO6fdV>$0hOc%2?ppm{EmjZ5-XQ~}SX zNK;>JJ=g50NNgB_6aVL0qe=y{LJUk}2znrGdSbF{Q~2QM0K}Y@?yysOdRlZ*m_?FM z%P10Ha59Li?I`U}^b@Y=of*2A@PWYNpR@%*V(il|k7~H4r}-^IC<%v@OWz@;Qq12X z-!^Nw#plJPbu{qX0)6Hg&&2f^$jVi5Er}q$tvNtzI3X2yg-?yu6(pNd$}(rDUuzrj zu>JQCr{ZA(#!yHYCZHwmSD1`tIVc6rzIZQE7dS&Z>)^zGbiiK*ke)z>Of8Shw~^YA zjRrg+P?yq>#FZfcQHau*9pfZSRj&z9JuKqA6ipc)v!43NjT8~7m2ny-^_hnGYHg`0 z#HcV*FaPYF&!!rFCg#Kl7B3>IPh4wqaL)}(Vf9F&)Bwrh=VY&c!#FIYkf8?qfrd{;7MAUoLwad z&!uCva{?5pL0JKEwAq2mY+<#iBZ(>AxRttmu6D@l{NuVGpG!^SHZ2$345*yu2yt>& zM^HnE2F}Wa)>+}OUDehAM0Jo#s>A-|1fM>g6d^DRy*`0aS}i0KqJ5%28Kj8BgM))a z?Hl-n+W_nS{(f0pTr4xwGqHta=z2qJBz%J5mM8Ew6cXo{b zGDc-i>Kx-cuu;#8?M;{o2Boec&pRM^I1EqA!CyEhG^`HL0DLNGCwdt6m7Br~XJ|V8 zix}^A-1hw=@+4lE#|~K;*Hkf-ucGj^MkCFFbRHLE?#1S?jCFEYmx$nh7|Cn3jWlY1 zQ`Y+p!$fjXL0gYBbx~GWsLS6m!fimrazwoVW4%`mV?9Fw)_ryq#kN3t_?U<`$hn7+ zZ)idW*ApCKto!@ixpQTXX^S5APgPx&VPp{T0U4;IGMEEPc#UkgdiK8)si`VGCitj* zM1k0QItetN_5cZa3V^` z>43nAq^KIdo5BKQo(LPBSVq&^qdJtRt z!XWC_)^^}|{P;0tchdINt5?hF%4%SVkqXu(P_MS#!>d$U37_{yNe9{{ za{ZW*AI0cE<=Kqv;Gp)U&UNbPDIvs*xEd2o0W$8L*zwpW~mEg(;L8ln?PCytW9J z!bp{|D4VwGcXRGDRqBk}vliO$8egy47ef{vA5l?UiY+A^l((IIq|VOvcKP(PPeNon zKZV^jF+Ew9SC>(7Q;|&CC91ufw{8XIi{~!|z870Bn1J2KR_m7qRAOfmLv=hj#Ffn2 z*8Ak2%n%!r`x?SfgV}(p1|g)tV3c8W0l6>YCM39Q%P{gfnk%_dju-&uo7B4 zicfKd)~}u-oEFK)pU63WcZ`?u9d+x|;H%>VGKY04xf>OIa6r7zfkuc%mK*>>1`NNz zrcXCY!PE3b6GMp|9E=A| zOl8f4u@%c`O`Tnc9mcKI0bFfUBny!BdOzjIzbjH0mDnjBqqB5nOeEZ5++uuq8_$@@vVjVd1msTrJUZ6s_x!{`{l)p7iDd2z3k5~ zlqZiLQh&G1&&`5xY&C{_3$>~0Py(F-i*{QAvq3hXGl(L+07R*I)}AOM^oIDw@`Lyq z>2!qq_B%F|OhG;H3gy5b+>*=&%`&e0Y_*|eO$biAqDjn$F)gEN9WzoTF1Un=oU)0u zl*l|}c|AiUi68`a4Z(xsL$Vv_BlRi_Mn~3}C%FJVy%AV=9)QmHEnFTo2rM;`RNyCO z0xG>dfNow6ap{=St0A@zhj9R;A%62M^Hnux2kP_wPADSvIijkn^BLd|^N0|kijP!c?IA*$7jxsb0&RO`w?8Ez{HF#;S)isWBrs9F$^x&G>7 z!`$K1NF#~w{pRqupXP{~r6Vyyt&4xE10|YAQ+Fc?{dPiBbv!&~gn}Tu<~8AO6X8|G zL6?LR(!OOu^4`vN*@bCt-n>yh{rKZ@=iZ$Vp^LiGW8B00+eNiC>Ol#%%QX+}E${CW zayZ~KMTlW`Zq5eFG0fN_E+zpUK73G~lkc^6uv;c6GdDXQCenMHo|z^iK*mQh;z`CP zgBUWKbg|#|Hxf!B(}@=GrVKI}i9ST7Mw&rf<4+J-smL&;JRW5DL&956#h>V|>K{fE z$D)>tp$gxsD{J74bMPpuP)=MANR!!N5JtjL#+)vhCCoLZQ7TGh*5Xka7MEtM(&Z6r z4{KyF7W^N}&HU%!sQKMi2*@`CI4?%#u;#HLV* zszoJ_47ISZSe6zR^1boKo25I^t^r0mYL>txa3=^V0bmNp3n2Q74a`icG(ukF1G^JK?$VJ+yjAR8V5{5m+_}$yzONV!N zF~E23ASIq>R4J`K`uHOZ+iv;di`zVOR_;G~Se~KsnqvYrJpZ)Zg3sTjD>_@EUcn^cOkJtyfhkJ)*5?gS6WesVzOo^aW=iA13DOz6$uNM-&<2^pE^(y;S7IO5yLQ8S_2cuwA7vB( zDFa@WNZ}>Qg)7qoGy(BDk)TlUx(py=ox)g?lX#a*BloQG8i3)Ajd4ZYM*U?XRgxi< z3)QHYy1$~Ql?QMXa=iL}$d*sR`#sVC`Ik9H3j z$Td*wqsnt=Y28LZnSH3#t1t<@ZF^t#?CxV*%uLS$Gn|Z6YI5Bv5jgCG;yZ?;$cgTy z63cr4Y$Mq!!*FTLFgDWIZ)tekFCVjBO;MrqTqK&BArhZHeNx{4#d~Ce?w8H&r+6-S zco3~|EVg&I6Ipk<#qwrAiV*+(eU3;2D*&c_)UC$8{mv+a7*1}R`GCMgePBzWZXF1_ zNLbaP=NHeNFM}hb5uBpoFTNbAA&Bt71Ys1YpgNVYoR>57P~%wiLJ=kbP&um5nCak| zpTScdAOJ)$J>$#*5CSleY1}E^v8|%UZsKr*UKl3os?d zwMk|#24;}9Fexw_jt)&*hG8afoGd0>u?x9|v`(y2)I|j&#hMAnSv;hMF_cnaEWSlZ z4sk+dVHU54p=2sL!$+d$xmr3MXqYb|n=&4DjHd$ZV1POSxnYX*i4N1rXPZyU4QwgB zD?iC+CoJHzjy{Q@%CzmMhYZ_s!~XzZ!e?J(v_VEYl>(pknsaEzp`6laV0iTCak2tt zV5B%4jP?oQWZT$&1?k$t_}8G7?ZK3_b0X|qP(lK~DsTwJBv!>zWy0W#fhu#rlHkp- zix4KZO*Pfop_0@g_r;#1l!y3<#G<`OHG$yF!E%TiG_YDj1=U38B1P@T%p4zhLS{A4 ze0=zXumnaq14RgtsnkNaf&o8=ovEqhAMVAzRFw$9-N8HBSV5hPmO#`tkM~ zdoh6fS4qOr6CP-)z%tUoZ* zh|Cb6(+~QMRNX(=4=xf-hY%7;d&nT^?c2A@B)D8(StX{nnfb!SWupB?w}SwxqfefP z^x(oiI@;4Mx)zDak1zy`=yiF`k9ba|DvT+D#O08I7pCHhc9F~`t9ZY`U(+W!Ldlmw zjU+X;7FejqrE`)_=f(Pl%$J_hUjx8_$(ku3>v0zxRL5_q7t3{V&S{YwH*O}b zb@kfya`Ez|a&>T}EH5piHVC?+8l`y%FMc41OeJx}SDn^@S!mP#=aYzpp85DWU_gB2 z87uQ&JU*^hd;^$u^YQSSa4Y#O7pK5UchZw}o@hSp8hke4(aHHKroZ;cBQ-GGHHa$5 zLD|^Q3?QXg%lZJ^EDzIgB)me2wFhkieKWZ1&%<2CXpFQv!Kvt%$Z^^EFu0>AZmwAVJ&uUr z278om?7T)q{~lHR)*?V(y&wq}B_d)gv-zMnkQlMW0NL`H_+{j8@4P4`cY(Az; zD-30>n7HVnVk)7>Np{Bi#3#%|W#I^8OB(D7Y_X9{-Ox_EjB9*ixC6SP{yBoh9UBRCDu@#u69@n)z3$ut%^1FlOjM^PnuB zMXh#;o=fipSvhg4#f}`4t@$N()C1<{o<4tCzPNikqsRD>!Mf&fg#Erp(_xv(&mZ#a z@6(0|%R!%v6M95K&{h`)8G)BmCUXrRct^Zzvp9`3%|~W=_Vj6aYvZj9nj@Tmjstg0 zXPHytJE!N0W39=UA>x~+wO5s5$9QQ2mSB0!)kr2Bvi`YF_)22J077bJ!wKrE;z82G zpgiA!;d#I<5nAmLsc^?3OW0Fe}27E%03=TPH8=pl1` zkDoj#pML&HxqIhMCP4N26!l3oOKpGHACYJKuYZ1(u^NO3TLWwJ-0Nk%_xkS!k~ku@ zt3e)rzz@W-e_YQq1`wq=-v$eH=Z)ykfhVVbVj$DLOVhPwC-fANV!@Y#foCIWAdpO@ zR4{lQMxhO#&ujtbN*9?`31|32ScE}ZwW&yfm$W3#b#Ro7co>mR!lm@t$A;2z(-Be1 z^a-CR@s0?c2zTnIn_()Th$aIzg-rSjj4|ee7i=#2Io;Ka5ebelm_~a_)i*<@chRlW zL=pw7r{ZzwEX3@@&v66OSXQ;4Q&?sTTP}H2x(?G6K$H`z1kLf}A+0TDgZhvtL z1J58FqYgkS@C_ON@Nbd++m41|BCf|o_-*B1zZc2Dg#-oZdmYzY0PYUY@Xq!MNESW;#n}VoCnG+m?(`G{5Rw{#6CTmOsy{#c-UGrx> zp#kmH{)gK`uY?5hbL=@w2boafLFW-O2RcpCs(Se_W*EZhYTyz_NHxD@65$aK)fiwf z2s`)WDU%I^)sNFUhGFV-mkn~$vZ}!&TubNXW@i@yWYfvYrB-Hj;#I>DKs~2Y8_vO( zjOgex8$rg@nEm3*yXEKa{j@k0W5aT`_a>28$EZ?!&@e!)=pW{9Q)g5Ga7kzn+jcbj zEgGxp4i#g)rV@U@l5|x~2jU-NM%+D@n30SA|_ zTw&nOWZ*#}x~&a|#0E#sZ30RpK^q+D3^+CAvM;@Lq04G!?HwZY9wTrp{L-m90OCzxiM#zJ z_SDT=H_F3D_v6lLFzO!9PR^#|61!ml9{sh4_x;pSWMh>fycE3J#oVr=$4Xv*rUn8v zegdcziuleXI+XxXCBpbQpLei`a?@sHJR{oT!Y2DHt?GdXf6Y$KlmQZ0l#4G3aq3Kg z_c=FV@IILls=zBuo-dJ(Gfmj9_HT%Kl}Saj@W_l!CW1dB zd5E$bD|8ate=>}9c+cN5cf2~D6&{~Y-DwA!(!K&f3FZJ29%Ynfiv6uIv@}*}?KK!x zm=!vz#mq1Atdf^ND@dd9c5H)AM7EOV#q-VZzn;zj03ZNKL_t)HlIivJN&t+-a4XPQ zk>@U6$TfOAfrvG7lG%0L6#&{{lJn-RTjfVT{Tslju_*!9c4($!+1>pu8U>< z+I5%(7{w*TNk~CZ!E3aGgyr0rVVD^j3A1y0q9FE7;18>T4*tPhj(Ke=ZM~#X!)PN? znlngx_U|lGG`Vf`Pvc!0MzSGsBb7xiMbW!~%dEY4tiY;C5SEh$!b&;$KyTNui@0hr z#Y3Gi2qVLc0o`Uj#D#_VNGw&94TGB?*$SnoTq?p9dHJm>6$vsZ*A&5n4H2M1W40!%n0QMHcusr@P)wu zRpd!Pup8n}V^eAuv$h}&8F5eWeyV=(iKskU9~cR3fuoPc(4h@WacU_YTQ#e>VK zRtSi$lQ3v9pz5cKr?$JZRc?OqMfv#i&&wuRuHvOG)~L)uwX#QWB!2xg_GbnP@N+XT z3a1|#LDV#Zs41ePvO_Yzx%j6fGoEPsp&hny&>Y}0sd%zg4R&>n_Od(?y2E(h)3DC` zrESuCSqs!~TKl?=@Nw};ZgY8>@D;Elv8c%Bqt?8^r1aF3Gur|+gejXFbwLilFHXL?~^K&P9L8!T02br-cc+LT2h(m<>Jpq#3xJ^7O*?NRrs$BxI zk!nggy(kHLh$CYC)}f>xBH>%wR@4Q62_|X^2$wJpP&-GGV~V^kaPuj#yZNP+GWW;d zP9DB^a9}HiJahu@W!3})ho z+bkfFDeO~J8BN4NV35i*@gkv>ayX3UBs2ZC4h2F|uW7AC;-%Q#xPHCN!%Uxl{#js> zsT0Sd_8`EA!*xKd7~j|R#Jh2m4W*jIabzf|Dpy`uTI^`o_i>4-7K#L%m@vgPJmn63fxs>710gx%q_k zo!Qn~4DsoZQx;8UCAy+ooQMchfA3;Py#M}txZ027{Ur(=l)d`XAklhK@2gvX_l!<@ zh;SuC;^GRWgv{>(fj&f2Nysp+4-|37hhq;4xR1w!!MZ_0kR&7Q zGonu*y!Od}{SK5!Jw~ikO_(jfORbY27qH^0Yb--hSLGN4b~4|UIz6V!v}05^GXzOZ zSGM0%5G3<#!1(=t`~AeYu3z7%bpc1iQ1&H;G=0eQnrM*VeAzE2?IobhZBi{fvL{dE_hsw0!8#iyfVk$cRn~vYcRFP6V*pzW3srnp!TS!DxIkX3-VV*uS z2Gu(|i)7XAQAMeewf7!Bc|_jS1dAnRnfGEeaebatV*Bah2{S-dadQyA!}ke{x`LXT z00(XAfz}KxbJ#&7IZI}jy82hVP2dU6lQLpE`$)vP#ZQ|Ijr(Ph*j3_mPIoA+V0^YJ|APu_l?xXxB?e|cJ*0@XWSkxPe~JmxitL&jCB93o+l z$RCr5avf$n238q40nu@d$m)3fT1iU4*l&ywRht9P6*(C=k%Ypb6suA(*(t!|yg7JYmtTDJi}K-zAJHs}4jB+I86GeA!kM9|N8+e;6vnQhPc?*5Cx~}gu`@#go=d+R(e`l~gMW%Og9ogtnPe*; z8C7D=!>9)NTzD`+858(akt<7!Q3tBNmDZAo3**ZOQ!w}uc(6?kW7*DU%|eB`z&sha zUj`-V*8oX(ZgrKU#>eHyKlyR_v+qro`Nbu0Z~i%S$HP3t`e;ep`1h1U3L{qV`E8p zpAnIc&Y%4JZ_6i4)5x)XG7d$+A9C1i!%UuQ>v=fh=z2OT*|RSLomwU3cL}M}gCUIk z`jNquMoKJI#5CUygchT35=NK#?3l*70opzC>6L&o=#(Q3f>}?WvI9i9ArTi+i0nzq zFA`oD^qHbTrOA9UV~43PSK4?OX4%4Q*q=c%X=TZSHAn*e-Z{95*l#;=#eUM+;sLBRc9KJOV`YglDX=K9d#%pEo$;1 zo;=SbzlydU;Nn4y?x20_xBc|Gl3u2f>5Q4VnyIp3)d&OHs194g2h;#lFoN|s-gLHa1X&>egCbKpzXVTdUqx$ms3uTR5C!^e!cNEGn#}sn4 zv(r+-Zp>mXR4(#tiE4gl4cySkW4 zqm6Ut?wxY|`bIf-{(Ls)RT?RsRCVIHN~^kJN$IS_QW_~)rAKH2{nqkV&+Uv{HqumZi6;)C~`4+Mv6uumK0sz;H(fI((r`i)hI#LXXrM?Nx=6 zCWH0LtmBo?)WaPs_E37L8b)M2GHkQNT>D^o8J!CfRDXJ950F+G>o~_ef$PvofD5Q0{Zr0M1U+WJ~KzkV(wM*=u;z6*kloiFg+oY^cdo*O1&r2wd>4okT^O*5_=YARmD7D#nCJ)vG>8P z3mrB$H_58qOxB4^C_YrRN~~(GAkQIUh%W80J^eu15L$aFVFHRdFlAKn9ccwqIe=rP z*R-M9(bm%PpTNVCBab2IF6p>>eK@-czY4vO|C+=4{OHjHY})`CN&%6f!Mz00B(jX< z$Br*z7uE7-@vhFZ41NizWrCtAX`YzdOb@$@4!&Sc|XJQRMbX+c4vE5*-&88MH_)m$8{?`iwZ! zg9i`dl{<))OlIqv_a2)G_)OW}+gAm$ua!evqWeuqL>l7Z1h?9NwkPg|&2yzUm%SPs z_lTQio{}r#Oeu#9hAgls(s<;0d7hVVWgdml5&~d-^;{V1ctDg6nA8x4T2#H-nH)AJ zgH_upSKxLlAZ$nDt=JY@%=?+oqt168aQ|=@HRMi}6`)4q<-)K0Ni$qCc$RY{l9I$| zc@q+tL4wFovx;gD3ie=#StO9lt`;D^#GRd*4MQ!Gjj4UG1jA^I3;&DZi%1&Fi=^qr zepy{zNm~-7haWs(^EFkA(vCurimS^ksfWQ{VCcJY>B{9xM37-f(n&OCU43&2HMO*| zUKaQ^2TS6Sf}4%iCJZYmX*tel=i_0+99mT$yU~%hn7)$#}Oc^_C0rM$@Lw;nYAiL<%jssxG9$TtEV};2wf%;K#G6eUP zMg!n?klO!59O-jgo0GCGo=~zGc$NiKdn4MEp*nu7M}#tDdr{Ki5@rS&72AaA@BMhqjMI$x36T6U?=M81}dE6FpD^laqL6 z+Cws{qkEoC(<3Bozbvni5as$)k0B2NBJCM8i?jXmwZ*xza^*s~cjrq|lL;{Z zhl@gYFxH>qo=ZH<2h8pA{K4Haq>k0~^Hu#*Tazx+V$M$DJ7=ho?%tmcV($yTI5p0s ztA9%Dt*3ni5*vn4f=I)tP>DocnmRxZ&TB})zG@!$PX-+_@T8gbIPkd{ny69qa$QTQ z!JZEDjI$g%C4RPyU%9IL%s(0&E{sLsYN+{8T>!%@$RC#p!AU$Cg`YhSJtDacQDVYWf0$|@mowg+3z(}{W@w` zW?cbcCrB8H;oj}K{PWgJyHk{FYoJa;Se{p5fUNOH2@%xAc5>&K$(5yYeQBYbyMDc# zp>{75t6H3%Pq6oY`QQInk)^^Z$AkffNXE}@-!5Ny>nr74whtg=GC*cgrUPC86Gk@- zLc9Di_9pDv){*;>C;*`JQZmUn&V%aJiC<}CNW_8Qu0U4kxi+SC+mHH+mXIn;W82cg z83Ti+(iWD$te9btU8t~mO)bRayD9G4o}15~XtTml+>q<5y;Ye*Z^1xhUA|2)5!>U% z<`xF!^GGa>QzUdoj&O-$k47DO&AMgIMk+bF#J-F&jg8bWp^0k}(r2>Q0>COCwxgr(Jd>yitYRc*bnhGQVyXy+p3zV@v_TzEEOV<-OW>zrEvY2@NrK}WWTq|j zSeZn9m)TLdge^GF`GC2|_dolz{PZV(Tc)w&Mn~2|76i_|-lSm_h(Z;WHX6`TH!Orj zY*4mJEc{jwuB5>rN3&GilcvLpD>wFaK&j5B>4VH>sG=U)QK&a?VdiL@3q0B^`KPWC z3>-$HaOffvQnt<3Ogr$vwKv}Zr|Z>ufk~^?+(DBpW&o>sFv}EojtrToGFAPqD05y& z@q2JUTm^15X!@81ni{@7*D%x=;t)duW<|9O%g~NwLQRFi(8r2XAC#FjlJ&)aTw26M zHU#TRW2MegfSrx8t{PNQ%{wcgST^=3~j;b2_zhQTDghFA_2`l>bnDCXMx%J zJsdnM2Up5DrXs%ePyRG+yVBKVTpFAr)d>4t+_9dfX%GPJ z#HDtT2N>X&>c+7|A&}ZxoGlAe6XlQbu>O;qH_FF9`+z!FiVxfx)|R$92c)<$PNqqq zLn5x-S5=iK!|On4iuYSw1K+3Yu^z?(-_A(rSlzvUFY5Y_-grAUkZwbk`ipDMrMY<2 z@K?ADNl@&tp-B&0Nn*%+5f0!-#+m@G1DN)z@hb*-8#`8S+JLEWcad6Z6Fqz+CgM^H zkBTC(09w@8kYmuwhiqqUBwp|6;iE?}dL8)=6^RlZ>@z&;#}0~1PE$|4?<8$?y|se` zB*|tf6l?`oxg=_Ct5hw+Imb88N@CTpp6LQiGs$RIkM|))JQNtBH{rBOx6Jm4b_3TW zQeYA()<(*>@vO#ch($j!DPXvMxd`KV%}-?s{<3w_i1|db73NWmCq(6Kdx(^nXN0z} ztJvMwPrh?!TK@E)9`-S`{nq~p|f5*RkrW7I|);FZaET?p-d?zm@V4g&9Edm4#< z^6XhA!-Ya!rYbyz8MV=#BdO5GQHg9h5$8;lXmcg<4)Y0nmzy&6!urK>5yM@Rr49n` z%G8zNO>dWvs__nkRFS82kvUXL2>Wm>U4S4{cwN0c({9IeRY23Hu}LJ6*Gx~5p=3l4 z!4Wa;qAQohkc{fk7J-Q{+>Iw49c*XB=%&i#nWN)bN0#9M!wsq;1GPF3s=pqvXj2(Y zH|?zrqZ;M555|+U;i#w#l)|cX%7Awe04!ilM7W+V^d2*p1A;7&(m@(_A=V`f)E0b!U4HQ6A0^N@ zaqLzC#J7OmIY*_o>ee>2r^K0|9b;C86qNoN%x&!^iZ$5{(!h3{f-{~cW9u9d{vanLjSk?(2bFfuE=6%cESa|_f3FaZ8?12De z9#DLXVJ;e6wKYAyW6HbOQ>Ic9xs{=|n22#IWdTv`q_IY!dr>Xn!dotTNP&?)_E+Yv zAV>$5l>tH>LSY+`qIRaTmq4UnUTON5bKu1J%0O`Rx94et94PN~h_C}-;Floj2G?_w zuOQjqIZ|7AO9>@aWMoYWpLN@yll`r-$l&dpG+=}4oB6A5;0xXl%s&A zD>G!6d95N|vY+1j4CAPsDa0*yw7Xf3g3{AtBfNAAh>utWuH&W$dodeAMvMj zJA%j>?+(zavQ)!LO`Z`a;3A2I)MNb0eNN555G@C&TFnJD5ly;4F-!_qVXPPU{@u{!ys$KGcAy7#w#SX zbAi=z9u-SGL;JJ{qo|T#Ku{s0J75ej>v@^747b<-bz?B&r5@r{uM@NS>O1eSv(Bx= zn0DEo$z6;m(E?7B#HnBaDJ|MaA%72GD4EGcFvBk5fa~o6lI09yBW)b;-Bau`B={!l z7UFSpEqwr1nNtr}#&H8*Lmk?;K5GwUS}aleCAdmIr*seoOb0D9U%F1jy{lG%TPqt2@|*u@5-giYzKLnRYSx^m_(lJPlr)lZQUiK zK<0FGW;HghW9mhRa(KzaeiKGvHLLI>lVN#!P z5ahw}BQkI*;jc#+Oy+~Z-P`kkEP^xoI`TtB-8q~Mq_OEVJ^DF=w>z5OXCvKpc#b_# znKE;lCsImSVMgn8fV7Q@tEV{}gpF!*Aj=L8_Lw6S>7ge3re!5VRMvJ!!cIze4JJ~b8Ewn zqAEf@4A3P`)P-=)ZFSiG(7>J2q;AUtjKG#b0Q`-b=Je&^2=$Yr(ouAzZl8Od5iR-h z0G=)qVk9F^L68CJfQbKqK&?!sVSU2X*FM>sBSIOy8TOhdB5#=F+&LJFfxu!PAT>g+ z+TzSX0%OQDqanCzn=#6q95{43$q8)n+k>EP`|Vp)L%V(@vZ*=;2!fY=p`1g(-idc@ zWT%vrV6u*_hdF??OT0?Hh??gCLmR?+{n{JJy%LEU`+Alq3K-}R;RFox)5H@OSL6|y zqYlQ}v;+XXdL1Cl&KX=Y0W1k+7~Yw%rD2UaM^EAy0Pc|~6&f>D)(#}KmpTMK5n-hX zk|y37?Pz?OHQ-8{73TCVU%phXUA=}&%hU!;W#>e8>UNc4*`5I{!bno~M$ z#5)ulWwX5agrHeQtcv8fwv^B-C^L~a>j87?HT~emR>UQdO1RC5~$`Agued-G~ zD}Btm1vTJu|@=e#ur@6hcbLreKertfgu#9B366F%}|d~Ienyy zNR+XZ0u`iuWva$pp*|u}Z^hV$#-B@&oO&_Kz$m&*2Gq2n+NyJWE|y(cS!L3A6%{xa zrjxOB09HV$zkg+ZC6*^0n8|0t&`2eK-~+NCp#4y~C~<{bMCD%)27$T$hcZ^}+3)@< zjCF|Wd{yQK2rda1Fk%Wg&o;#XI^NtVCm8YCBdg4L?ts@wR7-+E z#wsOo@6?oSU%7fcDLvljGUaMd_2v~i;#IuchPHAd?=n=7G6pOFN4z3=uo1j>g#O>z z+JSx$H{B(Jlza3pI@X>^Y#DK%#uu@Toa@`d-K`s_F_IVs+H#RThqGL`n# zno~9eB~zCBN~km3?t}pN4XYrktCY$}P+G~{kz#ZhU^6h3xDxZVLnQoT9|+4JUX0e! zFKfU6G;D_EX$CF>YRMdX(w-TBi_43p-2H+X1?D+nD%T(S0t3A~iFiuis;3H3L#Rrx zc!bnz@LJopKAb=YKJFdDJPczWz`C;a+LtC zaV#h4N8ok}2kPpzt6@S4j`7N~(gygHaWbI{=7xMO9?4vxgNnP{qyhD}3&TxA=NO^55JEFvDV=uuCAcEvH(m?vgm}CGBjdc-ZMuXnpB_=fI z@c`%1qrBPERED)oV-q&@6 zb>i0Cpez%|bCGCz45r(PvbfcpDXD-dHiGP<_;nPS(u^g>utU1xHB2!Ql}L^bx7A)| z>ST3gjk&!gqU>ZVV9aY*NS($r2f__F001BWNklG}%;`O64_ODz^fpJ@&^Nf7-veX!c z)Hr4^RfWJgs5I`y|1!Z*8Bn~M$N!ilzvFklLJAF2NOWMr`Ab08wlPF%{Z^y;(l(vJ zqc$A#$tNF|_uhLCG!L>A&s9WvdLC((Z`<{nsnJPWhG^1D+`_6U_~n=P5<==>`<;;D zbNzdNhWA=+UW2HvW>PXu^Ew*saQkiYU<)S$<$}`%VXe_W#dXN z5g?aaK5lNrp1R*23mD9Zs%dN(=K&SHK?b7QDWCIe=TMJMa2nPz%aha=Myb*BMMmv1 z%dTcf;scwstGg@Ez(bHzrd3)=Z{Zw?5MatKFsSXUccimQs#=b}d8lEg;LN6jWXm*1 zmAzaUGgWi__xc`FOui8p4uGcaF)#6ZTqw*6eB-7&Q^fAr!~T+UZ&eEOwu`I;6Z7Ja#Jw4QA3~@Z74lyMi38jdgRSXHQ`(h@OlisL~wZnc@?PYv( zy@I0Dqw{0J+HVhk{yK@2o4@6cNVGEJWY+IT`QWdr@?@ggnA0*@Jwu?8(Z-;KXs8ya zuFm5DyCUis&(NKUcLGvjHTMss5bx_Lh_i*sHZ2#i|~LwWv;g_7!136+&H zs#5OBBpMuwY_miSWGc-N9znp_A=?tN$k8btRjr5j-NxE)KO&X9-EvBB&}oAa^L3_^ z1U}*7wxU7iQEGVobp{C2(d{j~N%4TNN^FxWP0lovRC*}^rJc*+ zJe-0lJ_(5Kn+k2AsRBVYnnVDYfZ@;6HS8h1UH)&>J`zffv=des zL{U_)LoPC89}cEr8ODa1SkOxZ3k^qNEPQ8r0A^U4Cr6xmk9m>NQO7+7oFmluX9*dS z%WqYq6sp8$)=o-`ca+yxEy2 zT^>Anm`U4Xi2Q_WW~#0*qu>Pl{1l53V7iSrH+a7x>l9%$vFIyAboYf(?cl3N46zun zbbeHI=?HaNO%*y%mcrQ?QjS&8E+7`#NFIPkUeE{skYI_Eqs^NT68+OIk~W9HFMX*r zgWN^9+SBV-u9VA{u7r7g6>$%y(YiWA5V@~~Hm)+m5Q$}81)o#RLE{ro58^^3GHjo_ICFPPmK4l zLbMxuGs|?5WnD7(DR4gXRtVpW4(qplCc2ZTx_Z|^G&U$DW$ySc8|)nKiNQ%t|$IBA~k(LJ7?4>ouVZ$ zW6aZ&P9`AAKqDAV;R5$mDK%#5Fo;q}#>j}rIP=ikaAeC=RlMf%-}z%{JqO95@ff3# z*?Qm4a!N$ksYk9%$_x_9dPPKL^&!4*{p?{pMS`X%h_P~D0P--I(;Vhk?=V5? zUZze=E9GQVCxIWblIY%-_hQ@lCb1m?ezq-AFR=a2I&)f6sHXE5E@UbxY8Ip|qn6Cq zFdtmWAT#9vFAkyr?gp`Yypi?GFUu*J3VTG+-E#Xrul-e}J$?SHY_WJ%uUgwhso?A9 zcK1+I*o~`8vt=6F#W#XH0!~+Ei&t^1^l&#sS3oVVV4s4!VZZiSCr3ltnF(zr@nzi5 z0U|Dyur7$r)C+l6iY&dzq{$#4^td?m_R+HxWS9txA=i z20&v%IH<;2LJu<`Aegk-iOA3gdHbe@RMZ(9^E@_Jm-L?4tvRE;gA45`(!-_tS6Dgo zjX(NoIRZ{opCgf72e`y6eOaV^3iE(pj%J^YWCx}raI(qv;s>lnGmrAdpsihu~Wqke}|WkWEZrI0c9j$N{NlaoWsW z{}uACE?vG1?U9^xc*042>XDIS{`g!pnF^x{5H}*RU&2eBTK9AS5G1CE^{DP6Du-(> z5t8^CZ>lnL)L#bPxZW_P&xv?Pxo6hxXz=Al21O&0d=@0yaT&u73GjftATuW>@jex1 zN}|!IN-~3oLm0vG)_+8FdLLt0L<*nF=3S{R!#c{{P4N5w?ys>oKr*UE(QE)!3t<*9PtAPDM4X#e-W{@1*r;|8$v zejaSEEDn#n0bIl@9;odT1I97MwDys385;7j^H@hLrShWN1}UMG zTGp*3Ovqv^DXq&-JVKF1C;{eYG@oQilXjXLe%yO>FY3>rp>~V!<8gFL+=uLc@c#QB zl)LxuMx^&lK;wCB5$7mf4Wpf*_c?>%>)J!x(;hNMyida1!O3Cl78yrk4*9fUOfx`f z2*l&D!N3EuG?fT4xO*&Ek1Mdxvs&7?-`eVGrmQYsyi_)>T`m_cUnFA~4~udWGuT%P zc!Q_s>r(u``YQ&=bN1D8-z;rQ(uuI8oTkx0R~@M_-0w5{xl-sp5&Gfk=3OPpv>M&`=(?8K6XZ7zU3DmLrKxA3$B(*j!bO%3M`6 z5`SUj+Mx~+V{NMw`j|Q>6I+XpaVc&4ZIYRhWW6F4rybCX+yZHjiSfAE#(iGs^n!_P z<30y?R8O8hOGl2-^-4y!Q~Hz%UEjGn$Ma_(Ly6o+lxvjx@=6(gg0@^fzdl|iEzh`= zY@c+WHGb$W4|(~c?tmk;sulWr#E3m$KpbLQ;2a3EebEM;LOQzV)deEy=U9@zjDhXk z=PHxTCNGLF2bB^>f6RDR3;0%HLhYsJ44gT=E0}kAjcuA>QkUEL?Vb`cnsFsB2;9K!&Q10LhQfDPj zIMmyfRhn;!<`E+ta`=`~_hgW}-S1(7XFWHL|ECajNa$u9l2f4HklVk2dE(U;Om#KBq*CSlOPz6igx@cB}ImL z^=?HpH$NN6;woq{A;FzcJ=ou2#WZS$9HhsJ7-^x@SVmQ^UEjzCAWB>7 zw|jdy8}5vwcjKyNjpS^$3t-LiGMUOH|JopWPe*ur+a$MjQCE|=q9dgA8H9;lp$!E0 z`oeJ~{|RbV6{YPX4qZV6ttd-h30CUbZ(-y=d-6%tWpblHjpf!ZoGah`(?89`vy9_m zovJ>g-HX~&NFQfB=E0<4k_0qSlk3F8+>6UF#|tF6iG516ZGOj2rVS~8{0Qv~)99ri z!%)&y57_%^1nHg0WWLFPE!VDXWD_;L$;9&-TGFvfdvF4-#D{e>(#mVzS4gEdjUncU z1%nZ-N+^CsPYZ}ZyEj4(Z^#uKjf83>mJC?cV-F=Xv%sy$=JWEu_~nZn6YUz>HsOKJ%|c_^7u*FWujPP*^ShmK6#A9oGoAZ%2$|}UoR`` ztC3>jQ;D{A={(t`?rg(#`s?`Q zqkJx<$azMV#&sn!M9#lvFco8aa$G1cvt4{q}Qp!?++hP7VkG-`Ml6$L)&%j4%y z%GIN*WFkB(%dC}L#|HCI4PU!%zE}^4p*?+0 zrUFwvt!Z{vh7sM{>H-X6%xrTD6$V2%`d7(E^O$CJkrUHzz4Z?1-6lk8AHne0bS6Uf z$blkW4l^#6fBql;W%;ZB?V$YNi`$WeW;xhzO#umA@miVHb7PWv{mAUf2n+^oF|&f5 z+-@;Qj9o1<53jdjbVO&OF!6F+)gCQiOPNR#U|qD`UT z>8OSSRK+8LfgV=yO4Qbdyq4G(6>9T{5JzsLAS1@|;1nPM6au)UYl+CWJMOx@1(zsW z8Q95NCEmFeqT{FonNVB9j%Xh_cVC|M|ZtH$MLq73Jiw zwyfGhhCn)l_etzSJ#%MK9MC=-`VotRM9_-P@!!pN`()`|yH1$F32yLWmm8cbwWKe9 zQa5)hZD}j%KuR|lOhlz#+j{+|s*jilz`z$_Hu{B;2+^W8kV+6vuC3C@e#^D0vWowB z7>okv2F+^FxU&wKEXE)Pq;WAf8A&b?08HqeLM*x8Fh%m;SJ^}sF|Yk&jL0~Py9z4J zZruI?FNi(g9Zl4`N!0tvU0zy1F8MK%&ZJ!P$9(Pp$#k%Zox>6=y;UCZM`UPzgsV8E8|K1%!2)r*E^*cw+t4K352~@v% zM-9J)>BA2{Mq`kJKDG zK|Jd`^MOy;_Il@ZJ9ZJtxS7XW&=JvSXPJ_5EfJ@5HN9n|B(ri3RRn0_ety1WDu2m@ z_QkE1bmt$Vip^+ndgw3z)xRune(M`ml~NWGl#O$j%e$AZV8^xDz3)Ctz@C>Q_cvW% zD&P3ZyXE1tXJzBfcgxi`-^3{YMfu5lKhO3{E~;=E${5@E^J}O>wr`5;kya2H$u0wU zC^158@eQ;R*=1a6pp%xF5qQPYMgo9qSB^VWTGZ`6@#ijGW-u;ga7#Puh?x)uQW91B z*|y{nk2nC-SAS6#xbu9xn8Js~Bhzm_XV?&~A{jERV!n+4*?21B$uX|;TxG<{)mCf3 zuEXA#0#zlIRe}WWsRoUDFXBSaU}t3QAk@<}S9&1l8xade$(MPJ}U~%Uey+vh5%1Ujk$62Mm*Mc-+U`s z1lue=fuiwP-T32AzV;0U8PiTo#LxZtpQCC=I3G(ylWW;FRb%4 zOV^n@AdC{Aa(i=;Z&eEnDnn_wyQ7I&3?{5eTmzVK-BC?W zPaw_7C}lI)&pt0d_~8!|06c~IUwrlv@s_Fb-S2!iuiUt>#;Rhx#c5UoxzEFQK+ZM> z^A5q)X&f94?I+kbTY#|3JlHv;%=fbNXf6=$T_HO9|tHD&gzGb{u^Q*RL*NO9n+yhkWfB5hJdu*#jP86zrmMxLo zmGGEJjq;s8 z`40A08;SFbL9-7-;e9eE*ys7Yw41gjV1g_?8F#Rvn@f~kxpKaoJl*0AUf(Ehzx7VF zT_|U0#&m2~W+O5}=wvMr6Vv8Wa*n9mshbXwe*${t!Ujaedv%YC>)2eX&*vELN~>97 zbLTFs#S_jP?U+7eiZWTe`U?HrQ`4DnXMUBjJEwi94rW&k;sx_jg6SanZJh6`^uc{a z4P;ABm)FQBnjy)0p@pd2&>L2x9dRzhj#)6$Dh51dewnc-lBNV4QK$iE1ljGYedo-T z)$^J2a?Z{a)&p!1-}Yd|%}BO~m8)eoRiQPKhC*52)yAJb-x zu^ce_zrXt;%on7_be&g+O6_I`UB5$X_b24D+^vJu|DUKAUc3iD(l3(j< z3=Wt|Rcc0r`$p;gEV44j9hY6LZ6I`Rb}4TEJcKP#>kJ(|21`aDFi`UEfG1saOn~g zuY?iM+qx-seujOcBswKY&0t?lY_3yO!{Hq1%zBY;yHa|Y%|fma+d2Atx4cxX?Fc5gGyvA=x)qaI{B$2V>*qfTV#P6x*T47SGGB%NPn`|d;L%y?G& zM@M3W%Wd{)aPfEg1)MkleVmNJ8O+E$A7zA%>JG{c)4*I_E(Mo~xaEK%|7w&GAj zviMS^ZKSA3bIN3w%uV0;wa2$%2zSWS1I-&$0mT%^{SsbM(jf?{-CWz% zHk0vGe|~~5K*C3=001BWNklj7+z>#m4Cks+xt&(jbO#ar=Ovw8+@&0DtYjFYVD>uf#Ev)QW@k?N~AE;2E@ z#;Y`(!q|h+oGsUT6KJ;pQm2b}y&i3Y9TMeYct@lU5-@er+U!P?`c=L2IgAsms=ZGT#k{ z)xi(Q;IWwzscNv6>=6gkQA)HN+|#8nmhCuLWfmFs_-I5N?P|RAgo4sj8IVEEeObf! zq5o8S!kk#7eWQ-bj5YGEnij4Ml)(Ede`Khp11ZGlxx|W%=cK93Gt&1+gmZH(EABCB zd%c~^ef8M2Sm^}m#RW_e77YR9bS2Xa>SF@NkO@Ve=`m9atcs?-H8$Lz$C#bK9_i2U z>Rcq^8E*;o&8YhXXhoc(*jMYi&Ti*uD_)qvKl7nPq~QWbmP?&{J<0{_FkebJ!oHZ{ zd-1~gWNBh0%TGLmL_Gvz>T&?-8Jag`s`cv!J**NsOhG@rKzW(RLrIYV z%g%=rXY}U`b;W(d17T7b#vcw!%1a=njBdT@&BpLdC#urka;Y`XN}nnOG(`-F)I%Gm zIq&nV!amofH&O}hjM#pm&|KiYP|VUM-%bg zX|NtiQlwgk`q7MpAMiA9}_CLy0djDg$GO zVHBYP34HS02;;~84Y4hIhosbi@9g%+BX!`3bCFYjTUGs)OZm~<4{_*eB@ee013nw3 z9GA~cjcMX|ljPfjt3Fa!2F)x0k{5*3Ho2z1V=DIvgZu!A=pl}+Mw3IkEa0i4P=88# zS5!OQ<~|+N1ReT>FnHSsG(F~3zk?wH16s87q-o2PZ4YfL>y8aA^9U4G_*ut)sHt+zWfAVfrQT_!o-zM374 z56I6l2iw>0?hzQB!U0$yAQkDl5LWhRk)F{)uKrHFy%x}XJnPP2i!wR3%W<>C@q zB6!D2O{dSMc@K~h%|j{Y0FftLK&$lgw|$932ah8Lw%1gpS;NRZtX3bdM`a-KIz}H~ zp4N)%x2tr0xuZ#Pr$d5*q_IW{!C*4u{bx_gXLoM>5~6(`8b!1l7jsIX)>-r01CT}~ z$Z~YrapOSMwMHaUuj}M|4s$t%qHj>>ud(nGq)~lYbcm_C_>FnSHJDp@Xdl3a?Qyl~ zXga3H$F8g}4IerhTZaK*7O8}dWU-9#I8pKv7yht~eEq%u)Y`0&iEIYTfB!%HcfWU? z-+JraZ1w2_4o;6g>Ph9~_EUy>WLV!ep&21_L}lW1X(ulc)HTdtCWxLr5`lVvh`RLj z?6f9}LIn$#-{8VVIIKt9*8ON5wMw;W7e>}KlFCWom!0G+-hhpXbSjjvVJt?4?+28j zd?C3r-`WE*!_=MIEKDFBM0LeABjq0cty4A-o)PC^yNP1~vYw5ep}|hw;UONDTa3Et z+K_!8-09akB&G}(I}TpTMrK?yO+$312Vp81i{8j^7})kpgaj%6;X?Md=iArz%6Qpk zGG#o+#TRk>lVEL`t3FMMf0!e92NrErr7QfL$@h0JY?SR&7PcaFgt&TQoKnOjJ;A3a zglBJ1(tx&@{_&`T2gX7=O+XXh*&}Zlw-WdE!!qWqX^u=`j=07!5pof_A5;IH%{e(eHlbk;4KGh7u)8rY9YB`c`38uv#Lt9CZd$lq-)`+dK z71yyjwc~WdEbeo1fM`^;=32Klvb&E;iwKt~L554@on1Gb_VAdH0&%K7F0i@czBzuE zNPGW)`8y>aNS-+(2i(0x4L#t!fr$`VVnlj~S(qL{93)HhFh8BtO^_fVm>3!#&?N}- zr_$BjZ&hjCs#^|YG{ewgA^PeIER&Z7qsPWdUDehYSNZqJbMAp_{*Q( zFAK!FRA=6fQB7HiYF`D#X)gz+f!Ws&_ZWqYQJau0Do;!e7+^9&Hd*1i0o*x^cl+<2 zrmnuSzAy?EI#h#WVp^)x5rJ|;6{=C|kx#;ONI0KU&_zvy*Y55v??49f%3O*}ViiE+ zK(|LGnQ5f?dK0jx&vXm}&p~0vs(sd>5eW`D4{;s_;Q?Q*>(q{^K}v29vEuUz z1$Pn3V-84#hd2YTszC>)$(b^@b<0p*iw!*nSSh9jYvzk(8}->eU@243Pn(K8LX}~( zvq&&T3J)r#U4W@1GCA*nfS)5S zKYtty{)>bg-^)#|r&83cDi@!+4EJFwRIt-5=03Gh^?Io2awH|v((pJgq}WLt`TO_( z(=b(PMIw0U9m`lb>ZNEA-?pY?cPX9yQr-CGE_2UZr>JU+LGFVf?HHWdRRxAP5RVE5 z&={r?Z>zp~m(Y%Q5f+6?UCO1cX5-LBNyU8`QJhjM*Wnn=3+hX1^GiwRkb{9Kf+VXt zRL+x30w6F>npP<$2ZLKAN2Vkk8rw<==Asa#M;N~viS!d7CNjOC2e{6>qA$)q`0S)S z=WU|ymhAolc0gm$G6?xFn;Ww6aAGVhaLsEHZvf-Y+UvvEbBvu%f`{iX*Z4Tel- z{r#(lMP9q+B)8K@ZJ2sQh+^QpB2n@lqs^Gs(f`}sy*=M~UWa|(!0(42P6R*#q$r*w zo+(kHB}%esH&)wB?CGS_X>M|n$=}dvFIqRb>SR1^J?>>M+JB(wq!+#DMN_A#Jv~gF zO0neFmP}EUNP+}G5+Dc?XW(~0zn}GD(~vC7jnXDmUts_C-tT_j=Y5{F*0a`n*6Hc9 zMTDF>U>c5AfBt105NPa=Kk?)|Zps&hJj6b=tLl;BAVif_p$T}k?c>%5{p)ZOee7T1 zkNd6L$Acl1Y2i2#M<XUClCf^0XE7Cw9PRGLIFGBgsCh)mY-*jpkrVpU=VtMC}_A_)0{tnS3Y%prpd2*}5) zQOtL5>?uF#=%+bGFSDngv2x#}3o2;FSv>~Zo}mo@HOT;fL~WevZhBp}_miIW|JFGh zcRp6#2##~a#!x(-hu}Ls3|_n;I>wl~9D25x5a0_TTICIo=_iI%Y6wxv$^OjqUl{Hv z9ZX2NXcbV!Oz2IbPCZkU#9AzZ3&L3wl=tCg;mP~b*Jz<^cDB0LJO_o%coLEyQK^Sm8Q`3|&06 zc2Y1~K-~POM2n&_jLy=0#6f*embX&M+!4Si?g%#OZfZ6ea4je9rZNxsu|q$8(y{LA zPx@_rZs~E3`fN)NeeThTlGsWrn0&+_Ezcr~N@4pp#B#A{4&2&PsTc4ThWj5QVBlqp zt6@6UOs@KWL|AxXM;nIpBa5a!A}R#L@sBnZ;)+PM*$WI@=h;JKeIM&YFdO~3 z=iqvqQs*~A9>e7j6N2hBtH2rt$az?R%?t|m=~PuuPQhj#*v#sR^`hR%F3yVkKq0|$ z{H|n^IZ^nn$jB!P%lNQZDY=9O*^nnr9Q(Esw6vc33K8wXotioPZ#imrDV_c|u8hK? zctgS(iR*U=8U%Ko{tOob*Ir{Ii&O0u)MrDdt)8h03?&}n?u(G{e!GlqG)ucdNr6Ea(dT_Ro>of1X zJt=5Z6vk0~9=O!sv99+;g2n8A7&o>>oSnM_S`wufigHziMIn;OHSGGP_DCZz-Dm?; z4PqY41dTZF5Ut;rtgC)$%xDIkIKX}^?;71}B{fu8=ae#>DUdvOlfw5K{chHZk48v# z9~kZyf^_t9Nu;hL(oUw}epYD}$v5c78O`ya4KDU}YO^E%4zgQv0z-nqC9Gw8~vi~SjJ_ZN}_u6q^gC!W^%i8GX${k4M;PTza49pl(eA-4(@X|Ua@CnjD#;b#&t zt-j;5@0J8B^XGgM7Y?{~6d<`jz3EDm_eS^eX2`cBT@F|Jn65~q?FG9Vc^f{=*}Zf7 z?Mxewl)IS8K_7a7Gv6p!TUiQErvBhY7{bD|!p|L;gX?%%=Em-i3kp5obg`N3p|dIs zOVL^83xHb&I)E6Y2EaI*RQV%r_R#dLM<0E(OA7xz|NaE3iPJuR`2A)}9Z5ep5f@{j z7t%rzHRE&;py&@{4X%A|AqbFS&&813>$>`|bZ7>HaM)K244FB+`-j*N7fJ-8+W`2~{+#fQXf5GjbLwx4SpJVaA~U z4n5X;$K^tPR8FR^ZpCG`a21oXaa`Ii20k-`;w}SLcQU z>tmBGcwBkwtwMSpX;SBJrCm21Qme}QiwSRReW0HV+0jfRSK&&pj!f5!=mX-r=sd>Ge*Q z`2DR0V`BsFg`j-&uYTOop@#T7Get`3>}Hdq-91NN8ctPvw@XCLV%uHan?f*t#DKM3 z{MF5?ZR28B()x2fhpv>Y3&lhUMUk?u|A0Q~LGQ{8Jyz)OJsB@?FwfX&42b7y<&>_J zGbqULM!L=Cf901}4?J~W<4LCrCI_N66NY%deJtyCTy6B0p0OWi#>spC;GOfS5!GOb zYfH%8%*V>jr5MZY`*v6Fz54Fz^y{yU?#0G*qJ`4;fAH$+7hn9!>hJzX|8WT-ar$ug zy;8=0>o@H2tYAZF zhkw##oX+)eq<{$v_H|WyKDackx+jG{;%7KjFKiRii5CN8b+-a@F)E#LMq@rhBrb#I z4yF8EzFe$RL}ul(Q^J9?A=`;v$tWsul(^J6OZfBW&y5%TXnPJ~mA4W*Q*<^eke_#@ z>-E+0dNvKH^N0^i-q=N0J-7psN9>SlT|cjYweBBzWV_wlObLbB=0jEby{j5_>5EWrWcHj(Gv(Kp!am>W2 zO?9Ryx{8*(`tr-GGpFBLJ^!_@6%N$4FA>ki@yr63_Eu-#IKA4J&CD|YXdyx-j{C&M zQ_c0ap{>rm_PQxh2+^_Ci9(T%pDgVwVU=y7b#rz4ogdbh{j*olYp=bwdi8Xbj++RQW+DBa#1Xhq7no4K z{KYR$oIjVA$PU0in6iDK)$Qqb-dbHacXsuK7rwCi+AqCS*`aDoBtrJL2szfuc)LmU z?)P6?oqzlE>UaoO48+b9`_fmh-kAs-y64_KH%c}nDKT{T@bOU_K5XHku$(Y)U{J|M zlBz0(|Mp-1>#G;P@|Dqr0!ntIjn$_gePs3T{RjX4OaeHL_VXexz4AMLX{E-$mdJF*;EluHC#(t z)VN(eB%iGaaImedDE2@I?g4C2otlOO`d=yR_krTKH$ymE7W2gZRx;vv7(JOE*joS% zfb7622yYl*1Hm>ag=PXX!eR?CLnxo=Jvtp{kzG>fB8&zZ5OyODN~74XF$>f;Tjz2n z@1#H6x=fdoP0W-~OjkzGsbku+v2X@bg+Cr?<@NU2)2r8C`$62MZQerUV~;#Gg)son z`01bnuwu2$(u;A|Be|~LPY2q~hIcq|^~mbM46NDZ0`LpZe{uEY=f5FeRiNQz@%V|mtKE}uYhqT)JDtz!%O(NprqVb0R;4Ad7t&iUrApY~a>_6b zX<>5ZC&qyk=xxrW=mF$(zyl32ArL>0Uc(}d=rg$U)N z>QlTQho&dnTni2nAvMA8`Z*grgj8(^BDh#o?#u}SY2ynb*x!U(M?mzO5QAvFtbIj4pn*OaITU28l?gmAEwmL(l-Lx9fxMD zHG*Nb*t*ebutd#-O1f6WR_&`piAwd&w<`s>(H0(&-sok-rHq%D@6zr1bv-e3qgq&# z*OzEOsE<7I@Nh_k6y>F}&<46Ei+k8fw4Gc%_v{xU(tWEp&%8CyRla8&wOu9#41&}Y zW$z){rHJ7aW*FsK^yQIa@J*yg9x8S`5qCJ-(@w#68`ldFI+{+EOK5h4ZIW!Xy1e(v zw!qBC-y z_~jyH9~oG%!3_s$$gmfrn@Dj@pIh}tHyv?@+P+AaELR{R7|z5fKEKxX%@#3J528Nb zs*2&2eG1#spxaph?*-4pf4NdimbVUg7)TU8ptM*+5{v#if*FP6>crP7^3 zAV)(OHjhIx(gBFyeAVBlc)G#DQ(~W<3AeCCA#(s zV{Qb66Hwj}w&Pm;lUzG$K;KoTv4Hr)|Lymye)z^@Vsn{ML1cc<0-?!st8iO(trO{q zC-1v|^}|=bU)yoR#z?mYC;xdJR1(|>XNQk$l=8<;xo|LQLk7UEPZV~|2wnK>9`!xW zq_@KOrVcppgN@nSH}1r#*s6vN2Ep%+Vz&%r%HY(ZPd?Nt<;^!*0*CyNsT5&~ zT4sr6YOEOn_0;r)EbB?4E0@`2Dy5ChC&HrL;SdYw3K4tsuey6H+ey1E&@+h2(eoMr zqojzFQzo6mRcViApfwf(R3%5yq_|H3s#N#Mh-%AozB-Beu;HFvV+=`GAPyXpG;_ zb-mT8h_27~#QG()zE_QegNd@+`!f?HF3kW$72OO_rbKhh+4jGp3YL{864GgZliOxa z*NEUA9ja&GWQwuMZdRxLSR7*RiddyrojLPn^8V)Pi(h=9q8Sg(>=Pm_W12-@3R$)H zXTf2H4{}jj3Oz!TW_I5LiQvYHGfheG{nbe5TEuZV5o>>I07tZ=e>PKvtVsM&(4Y6h zmL&i`GbLajj8NI0$c-gQ11AkGN>8RI*zAg-Rni2t;zX;?^XK1>u-=+o%@8*j&7!CY zV3A-!A^FxpSDMY?`>YP>m1~z)?_`NT6`|~wY8E-hk>f9!9*&(Tbf#;l5F~*TcC`ei!8_otAVL9&uN7(nky&)Caa_{Qlol_om1I#o!`IJINH!?7Z- zTP1^D`tbVd^*7!uT)Xp zpMLFh3i{^iD=+=Z>_|#nU7%~lWpE?<(qz8S$GVO+o4pFct#)@K+C#_VES?sCDaL6dOq7t6_oT{U`J5?9GV!*r%!rn)0Vu*oBWinEj29X0e_q$9hYweEU0p z)Q%B9Sbh58hnu|-RP27h91^r@MIp(dhc&n!+SkS@3d-B7`|bZ%8){M=qW8>E37cF^Wl&kuNV5Jgo= zFugy0s;M@$J`d)TJCVYFs|hAMNy^&(;Yv*qBB6fm(S2HJb)vgMNVH?63~_sxQf#jz z+7WJxfX%pxiMLrH6NGcF1~j5vY#TLJ7$K)%apNR!D)lo(m|6={68)hs5&3Pl8E)K$1K zUC4Bz?84JeKhwnHbG7I)(*i7!hYKz5DVLS~ft&-03~5{RA~|WI56r01c>*M*+Wlnh zv3oxoQXg1-aQ3~`yBXkYO!TFy2g(Gtg$W&wPtQMxaolWauiblM z>l?SqxK1aTbf}(9MACcc55M@;ugwg(lx)X}H!?UXVskBepjL=yVr=H`i!7#Rfmf?a2Cwq z-VCSb&YfSKJ9&Q_V#Pt5Eg5kZoP<;OgyV%q5L*P%y?nM3EVz6~q(wDpVkC`lSa9;W z5zjU26aWAq07*naRMdy=?-Lz}nPOjYcTU{eE7mU4!s~CmQM5#1e8)K~;P9=(3AwKrhXGdqED^P0H zS3sE0zxbuqALUZ$A^ElC;*Qq-L%o4o@o{e$zgn@Q(VoCqi(0&ukJ5h zUnM>1?2MGhN^kd^CSKNe-UJ-Arh>_&D%D?;gwfK%WHYe_pfx;zT%7EPt$WggDBQ%{ zmUOk?WQKCG3t0|=5mf+h_HHLqdsxT|LBU`zTOh(R#u;Bk@?I){Rn`NKKekk2?stnD z+?axX{#=FI&b*y27;*O<4wgEgL{V+`PYqL5cow~PP(654BPMj` ziPElLZqTsL3$=RJ@g7nmHPLH%2uH0bfW@{21HBS4>(v47dO*1{? zYk%ceCNufMJ2`h7$o^JILOjT4%5)74;OH}~^s;k4jIic&K)7~o--yRYLbMrwH9&j6 zw+h1ts#!&pQ1+#7O;_vgh=%%LGlR|DCCmrrDXvgv7WZMoMRr4;>nHlM;V9LI=qM42u^=py>PFXIi$ik(J@vIOt#Iw=bKtSO)3|cu zyxCTCt?0z0Es~;@X`I@#JeTW5$$`|nF_YlJCZ63mo_d_^Z`qaNzV--0cpLlUvca3X z3idw6?X(v0b^Qjl$K&fPy_R_e5fjh7)kk+WfqHj8rZW@KEkk&VardEAn|m{1Otv_( zNmYo~E(*5x5tmtN#-Q8LyJ`SoELid&jhdZ^q3;25EaI_=PS0J9sM#ixFrw+a&djS{ zV^XUppMD~H*xu?=EZovuIY=`XuyD^&9vdV*PA18C%L9-k3w;t36ghiX<3`+dqmMV+ zV^6@CLT9(*R>Q*O(RZ~OXRgdXhIs-Y)qO0;lPMe#0oIL>*_DZk+pS&YiSJFtdV1cpd`HnxRO zY9d78PB^f*I2LrUb_`(+a27`6IqQ4-*mZaA@A~s4*IsLaJ^tB;hO-YyF_GF~Bb*1K zN)Mh!JOBcldDtbwStAd1FMU{FAGC+i$e%7VW2cXGtPc}SP>&!h`Whc1rSxR6LziR! zQd}9dB&g|9&wnpeuewbDy|!c<7l93 zwBmZiQ+Ev-?xF5A3f3&iTF7_g{Wx^}X-> z>1^-TWZrvB{+XJ2FD z1AMni&{m;h(-3i^Y2@>vv9}K*yH{H5RxE$aE)fo2>}K5bjaOe=edUE0;!f`9IpKvU z#~8aX-N=gE?rKv53|#8h5LTauyL)!e@S19boyX}&0WL3_W>MsohjP6r zmKm6K8XLoWgtAt|dsq)e)k6meq3#Tob>BRAzW>=Dlsf-wkZis~rgK4GJGufSeR87z zV{vb`A4q@N%>vEvx4+p8uzvjgo zwc-r9j;Uf3d>z*t$K^!bR)|Io5+N3UW$^DBE+$s(<;Xly|MgP@h>37d$}>aqU;6sjE5G#K>U)3sa_)|E6}hcEcM30`rJ4me5}V!q8TJsAiG?$Y&`=Bw&}0

>mEZG+wDc!IM;bLW%aQS^1Xm5AzZp5NU9C$9U?Nhp^&>k)@I^VS- zPq%ZrA2`u|IC1N#7Zw2o&M}DINKaeaufSN`nrGY2#XdzJ=E$CywjLq+BEFKyOlj{q z#&#?B90Fwu7!ypeqNq>QN74oCNUl+bq9nuz2aMGY&m7&WbHZ=lE?&*HeoBQq5Tff5 zu7mm%Hc@1MK0fLlfzXabOVLV`&D3%kY&_1Nt0GAcBc&pqfb0A{zGRTDp~ylgKAh55{fx;gE46 zj&8O&$i6o>$%(>h&RsY^1OO){oEq*Ya@FiBw24^a@_Hu`C4yEFJB$H=9mxE#HsgT5 zHkXj}##P4<5-u!r9L`4+QfublGUVY{J@(k3Ke&a$z2rPuKT55O^Q9q2Wb_ zAsj$h5Wm(50^z`O?KDOVYtFPLq~!riqwC-2K9KBK|L(Kj5fcB}r4Ld#voz20H-geZ z&%puY6*7iOBE?1QyX?ABi&lptmV2+CSY3R-O^zNcJch{WpZiDFPdg!Qe@c2BV_)Gb zM>g6fE>XC1wE(=hkZp(;H6RfYk?n-gsk?PJKW7nO3oB&T|h$(>09HoK+sHjj%J^*G>8`{~$*g}K4xyj!g5;)09% z<$pxL9=G6gp3pgCZs>OiZQ$B>BK6GMXR4?F;M6u-<6c81&@A)<4*V7IWT_ue=n(^% zkP}ry&dk*5h`}tb9u&9`4hmg5rgLjz^7-!X&O1iF&vo6@p-6}XRYM@XXLV9v`1ssI zFQd<>NowP5b=;j390v%)7Dwlml=mvo)zR8a}B!kSMcr2m=u!bgCOG7Mrk z>ZFo^A!w%$-hsf3759=p7wN997c3}BLlTP2nWvw4I*0MaR#}O!kh8sWaaMw7-+g;@ zFoI^`F1UURpYexUG&SLfS3Mh%mZ)nGw~6ZdG5TlYI1=IFma#PoA48+n4uowqA=DXNkby%jLAtJcZuUvjKI_PMIbS{r`v!?3gYP{Y_G1g#0i`Ss^P{Bq%(yZbssv?i)M zmwbF5WQYz36yqwvVHg;B*5&iH2Jgiiz;zUGnvJR*n<-Xd*NBmfets9VC^Uxc>fC!d zrL(6T&DCSadOFQ6k=_7Jo-xEW`)1-8Q-Iw;fho5F)&PUU76Jql?(Z{q%~oUCtsitB zyMoC`&g++6=)Q<2MlpbI{f?+=@j#;!Sx6`^)^E5!=I$Jz7jz zJC-#;q*@(`Lo*eC)J{G*aF{5E?g9JgHTcLCT|)T-WRRk*%r zE{szd%tFV@AP@>7yPIHi^2fise>#8X{qFnLI$0dKjyF?^q}(h?VnSK!81^|FEQPzr zjoAU-O)x#;6YQ-Comt}G2Wqz7aDZ${rAr<8fuR%9Hd6|p#N6P z$Kv3~byjor%{dZeO97RgCPI&uS&pNq?sKWtqOOF-c?V331T)YyBcc2bUX*xTszyCD zw^93U$EojY(h8^X&tYuSH`mt+@i_O@$4nZ)wH!Xp#Q}9mOk8eU&Y|XoaMlF1{_bw) z9BTqvUvu}lX{RM@%?x1)S62Az!aMJWu{)wMBq1JtDbQhkw!*XJ@2*(;h@c4nt|>h# z|CAc;g5fx9ts1Wez{AGfb>j3zNm3sHkNsxW+lZw55mewvpE2Ha$Ov^3Wwq6+%w@8aH~aoezx>NX zlzf@@7u{(i&R_l2Us=7B#s0Vcv){~ky2N0)r)K95v{50p>wrngqaccN34Q6J{!;}G z1W$DsmW4{(w6RgL9s1u5^Ww_WDP_sE!z2Wk;=d$!h+eRunw7c6`#(qhF!6l_Z#3Z* z$?C0Fe*s`Vbs=h;hic=`xYOMP^YJxv&jMhLPUq{C-+%mi1F`C)>?tc~+uG9&itb_w zjin2}&{00YjP!|*+)-z|-N?)lDlM&qe{`Q3CDQH0kaDybF z%c4TX)#C&rD$>dO9+B1OCH=mU$6dfwRKi2Wxu1IU(bc=fGreCe1rzF#;_#pU!t+IA z9$$U87`ET|yMJeFMwb)$P~wnTV|6KgZwMd*>EJIME_hk4W2d`axVO|h+>1F!W|2vQ z-~f6iL0oQPwIaIE(#eQ5_)0e;8Sae%3|<@S^dY#ci6~NP^2AhXzmxOez~J*T6H_eu zf}?mX7wF`gSiG+}lKY0@dtV#Iocoxl>hUOa9tNp*{p0chc&gETEqta+>IxGe=AY`KA^5>MaQrTV+zG<@kR9Sb_aWK06}RUy0PH?>dWdh9Lu zx3+8ySj_k*i{7xM-NDME(~Spr+H%D-+Wp$-82t@l_B#clGyV`|o$-hsyyiBYm}5;) z>+`;Aczx|WWPNe(HBqzu(H2qCKMy92B@IX;a-Iw4Sf+#XZ@$%XIE6h0ls-jal-{I9 z0@uaz{CZh?kYcG1*xWz2YL|SMD#kf#OyaJHKFaffA~el#K1zg-NLba0H!G@1u!vjD z&ZX#reXD;H*I$o!7(YWIHwsm4R%nU1-Clz%L|LCA)d@kKm1BM3Fz$x|aSu!BO%JRF zD~0+-_NEWxSbyg?|C7}t4?Mj38-Md}uD<^De`EF36Hlz(ZKH;_-a4H>?%e7h{o{W; zP?bqz+;G*f3PWDDgR#&svB#25|B3h~NfwNw6j;??!IN7Db5%uP5*eRAs5rl`}{GH+)X?`etu0z zcfaFY5Wl|GLl715qJ7x#u;$Hhq|gFVSSCd@6(?yGr)Gj#Srz8}=@Pdhn9;lH zFFlJ>cH9vWGo?6MtBpieE_SyWm>f#y# zH+l=x!jn%v)%JH!tzLQKhpYeUzy2>*-}vVLHK|2Rpf3;o4Y9Qi6U10?ag%8x8{oLM2d+Bb@BYrfhx+=q=1B8P7uMu#~{a)&=P+$qVn6(LXbc8$V02suf0B;XSacIaX`$W+LOv9 zPtO5RYu^;Y8jF0uoR!?2-_4_`Bpe`cE_S2U&8RDpCV6ybcoC+S&GewHYDy5bBeEg@ z&S}?dWh=v7*{u$A9ga)I6*E6IT5H=m0H(0d)B>kIsJW#d3nvSW!Dq>;&Q%jPmg0Z1 zt@U1P%VC7|U`qS9zV*#{8K=hV>ctU1b+MWqH64Kl^Tbtvno^rrvh#fqk?lKpw6Gm{ zh^YdQYXUTmF(beUj@?*{?Y12eNP;-iMztYKNGtO7yUBYtZhGoui7|!az}dyheRBGX zGqXn#U6#v9gzuvdR|+YqgM;Cqn96W0EVj1mILHt#;K3FmGOLLki)z-Rd$ovwlK^xwF=#M=v{wqiU|a>juW(rt0CJ8XT3NSb(xZHU^V%yups z3Ha8c%d936cLMgfe zl=_RwSnS}Qk{^85Z$!SY1&c|711=_&F784O@DDOCTxsR~=w}|wEAZaxl~=#N`jeNx zn}~iJ$^W_I)TswoFMj#UtIs_0*c99~|Jz+-@{0Xre9ManKim#SMs zL;x|hHFvMZbrMtkMlzw5-bPyU`h;UvUpeny|IGKqhPv>;#!zm^S4mNwgxc(2bt zQ)Ww``=9_u6Yk1~>zbl9c;^_r)bu)Bd`{QrfkOseMTH&T_^01keWnoY`|^vL9mF3l ziPKOSK3$FgRe}3wB75U-CdBTw@P|><&x=Kco=wLK)MZ?qUA_G3E0v|PuXbuh1k9x3 zx$~z!L_IN&_g4SQ@BX9Jzy9^VI(u49oEVB#Bb^;bEgP%H?b+1&GR0x4HC_{ z2J2*Gs_KUn?V@BLm6YER@kz;ZldF`^^e9cb76+8Jc^g*c;hf9I8#2iQ19 zBN*`_Uv8B3dt{t6s1OV32%2b91dajpR={G!+bJP*BS7!xvWc)NTvhvpuLR`n+Oh%W zPPHHyX2(F=b6BNOn=Uq#?q4Dd#6|6oBkk4x_{I{)>S68c8AK=l>4AkRr&fjR<@n;# z$KW}U1Hsd3N?O(pI#%$OmM@)cU)MbQXd zh?jGir1;LC{_*P7*KMMHXtiC`gnDDf0`^pb``fEpqAeZDy$2bP!|&j|Yq*ZstoUcI z=I&Y_bUd6>uHtr{5x5&-?*d$0__VzyfAGUsXJ71{`cytcLq#wFP(4v%jI$B4$=HPv zWpk-6g1St?Cj2IA#@@$VD#YIJ{_gMgUI=f8`0E8giykxiZ!mY)fy{;|J8xMFFg6f) z<+UG--sRHSZnjx`iMR=Y+j~cPu7dal{w1hzR`)GLTg7I!n#N)bh<38Pqtcz-W8I$9 zH-Je1r6cVNz>_~nO!SZL!3ixbOw}0r0^=+`Y!NZKl35V~h6hF@qDe=8OsCw7=qH&V zgsm4dss!#bWF><1MWiU$jWSI1)_-M){z?hA2n+H4^y5cicsGJ^LoOYL-H!@kc4%$% z(cY!>zj6I-r+@09UOZHd7i9Vv+)^#pg!JP#I}hGl-Dl5jMR#STr@Uir6s+6}@4dhJ z_P4*a`lsLgg9v%w>a&l3X4tc|_#gc6_1a7c{wW6q&dLqJg zB=5xSaKl-Bh3xxgH_6Db{|u@-tFmR9x!p3eWxX)C6-VHxW*t^!WbJ_E+k$A}H;Sz7 zB#Lt41UWiX?ZmlQVKa@j>WhFXkz^qtN^>*6(Cv~l_Sz+5>{k5_QS4Ql&B}F&OPHAG zhWa#%fPNztdq8_F=2z@`Wim}Rk*fMUOZV>MbNtGeeqq1>LHvKj*3WX_*xE)E4io0i zhbCWy4`>~@)yIU==K*eiotu^2aHOPI?Tt$0roZr@hr=U{ZMFLQ|LuR%ZoI{6z5eM#?i9xF;E|vcG zA`XI2eNhsK)tnxxPhB+nj~&d|m7ae5$<@F4uYG;>zyALJF&Q^MD`63ti`Z;D!ETV$ zUtn{vSq%&acb(`Rhn^4Xi(J^LQSyDPuYcLGel7{b+;RH!8>`KK@ZbIYAGy!c-pqkC zFNCwcQ9+8Os~b3D(3d!vV>q+(mETZP;pZu={Lkcz-$pfnNU zs-^iCV;h6&!<7X1YO5-Q1?DrcQTD&UaW_4DOd3N~KiVj|0=GK!BS%wW`uHO!|B^3= zidn-LDh5n185Mp2qg>;9ObG0!9b*ap@gDuO8~%ddJ|0uhpc~OcJX@7J#%W8Bb)e&{fpM#EB7?y{B;%bNFgRw$;G5x+aAr0=QD( zUCc;gM?jg_CBUvPe2!6F@My{3eWSqkUG-us)G1z)d4WYgHaFr~B!f(t?b=ny>~_j7 ztG9G?*6qofiMyfNy^OTGnFo}+e*W2~2RdW$t$z%*Kl=fq?*ACFe_>OuXoP?6_rLtk zKX=QVpx#Wqh)f254>BQ=E^;@e|KwZ$<(wUPUndoHm0Euz&LjELMo=Og7nnf~5QNkS z#|jjOg&fMbyhBKx9Kfzxn<|!7(qg>bj<{}8cDtUA#U#Nojc9N!Tt^*!>1q&QRXTA> z{SIyu)@uol5bFvT(ybN=Hwr$y&MDpkKprDtOI4%uEuu4%M5NASVRcBi`GY_F!`0g@ z_2mFC75wbSLM{uCe){KycmDY+fA(iDcTPpWf6+$%zqlXQx|-rE$Ps&$TmS$DeMv+? zRPR{ztMF^-^w$a@QNmj?#r>!9)TcW^nn_`D2nO_7Sw$jcCQZbH;OJWr>44mWl`_JS zm?9L_6+ya_C40;(2r4s-N-;Lxrdttl6ZXO+5FtsJP%A-+B4T_Y@rbWySU}uR4W>3O zR;F8)w90`P(N5iHjQ|G>#R2Ob={4QLDId zZjW$XRYIkOggg*y1ZS_1k5@^)KsjSjjgY1$Rx9c4tL5x(_STR|pWL zo!|1*sr$#l&OtmO-nH*D?n!SI|Nrb$i#QzJwx=W1pV#A)OIU$b;WTwWuU@Z!Y`PX? zChX!!MFwR1x}1nn_fnRyNXwJ=JutR4E6YP$3elEM6(?~(u21{+RhU0yTAdwlm8+UE z;uBYIC$sxYay?oC>rIIhoj+J6(NgLsBB(7!&-tD>-iKJ@mJFb`cBI_%-`!rlpRLdK z#l+e5%o|6_#6FxU=6c_)D5nx%bjZKRZ_0J0c!y%V4v`(k@UyNCr;&ffe;6PB-oN5L z{DqGJ0wodLt=ienjIdGL+&Zc?Yv+2zt^&-(49YCx zp82WL?v-d-e6n%TA<#6o#(_KA;bRhXYvYf<``z@cvlGv6h5Pe*d?Gp48kj^DZt`vk z1`k(X^ud!^(2JG7SuB_QKcN{E^D7a;_zWWk##wdD?}SDwB=sn z3>PGhw&NyjFSs7Uy49-lcH;9^iu}!VwZj!=+f9e62N6tMtCeA76%pjr(YHcz*u)S@ zjV}G{*~BViv4z9~?KEpwjM2{+QM+y@i++rEJ|pz~^<$z_6ZI1f>nV3c_|GeGeWD2q zP>C1Edu>DV>0*{7OCC{{sDE^kg}9pCA!0;YNWaCKpR%psnQeuc9<0hBi((JCL=S30 z1oSP6@+J*FdlUbTD2biiJ9c&eB%wEW9-A03i$hI@Orp|GVtbsk5y-gX`V}ku;@H*ypLc?4Xmfnx32V4Z!~mfevS6Pqj_Gi* zPAu3eozRgkML?AJDI&60QbBZfJ;J(vy{$#c+7a}}N5@3KyYE=riToyZd;;AaU2WPz)r&8VSg)cl`iJ$vNk~zdF>u1iK87DNegkZX375_jG2#v1}*CKKej~CaEB4;f0s{%F@%f%&ajxe^g z+TTvuk4K`XY1?lW8Baa)?CQDaJ{Qrqy>SFc2F~srRt6|6(d`gbLYcQwL*3{qo`NOR z+V4c|&-3w#Cag7J7XoNEi*%t21)ZI!)&eWKRD4kg#2k)uPj;wF6~2|EAV|+knixGw zWo2CNP?q%*l_25{62WmWBj>n-G9zeH@S=9D+*Acb-1i*2Z`w!T)#FcmHX;?ZN;EEu5AooF-^7%Qf1mg={i`)Qp}J0>o$w+RkR8PB&-3wf zL0I&TwYU9zI?MydQ{qGZ&Y>kWedH%CJu(D1oGb2f<7PH2ZKPKbePc36 zd0$G6r%0>T(u*}*Md&%Y9@i5kaIv(q*v)Vet5jU>__3qKtJxOr-i)8=e(g>3;Hd{{ zM}dxvrg5j5SdO^Q#DdorzoUL}FV>?mxDv}e0kU%bBy)q~FZ0tb4S#a~{@+C;X3EDn RjwJv9002ovPDHLkV1gr~(n0_L diff --git a/favicon.ico b/favicon.ico index afab23650b11877e46d61a1a5597ce091940c855..55716783dc4d7f20a5994a2217840c7150ba7ae1 100644 GIT binary patch literal 1406 zcmeH`-A~d{9L66jjh8HrZ=x+}4)w%(oTtgSNJidJnY zw*~)!jlAH-W_e?qP3#IbVvIpvv2nxbAn00~SKV~s=j?NKo@eLWe76IXki}ww>@*CO z0+j$HL{dRO%=g<-!LA0Uibv#wo0r1}uSNhqR|p|b1d*@-;(7@hL{dn^GBh>HA(JQ| zmnw-WD2UB+HB<@>)XElUHEn3s>Y&#d&~7lI!`OvRqY2$6Gb}y5L@O+2EBgEFun#z3 zw>dCqbAocXKsjBw;JS#ROK!L?4`bMU6(d)!;rhr8+`4%Ow~6=ej$-uw7#@s`6DKe+ zK8c5ukMQL26sDd|WBS>1ynNw-$MXtrUcbf6yBW;Rc;THTdcFAY-Upx02fyEsxw(%B z1Ok}>2~h=qmESX}&qrLRj^UjBxamG4+x4IvZ?VQuXP!r?F?;Rqs;2%^y_Vn1Vu z#bSuZ<47bDNG6j=rBYb`wT^T;jZ7wkY&MHrF1OwIclw^LtvcoMliqNB}CIyps%$!ezer-9Oap;7-kf@GITCH5l z*;{WZ6zBN#-Ac(W+R?${g;sWxQqfRlEJn&Cu%7Nb+dQzD9ilp6qfG7G%#^r<$$)w* dQe8C%*++Ue@|g`zoiwx;PJa*kU%qDk?Qc2)yP5z1 literal 1150 zcmbu7X-pJW5XTq71{49?D3TDaE^=51^di=NSV=EStJDNFKrIj;90Haq6$;x3OF86{ zU21oit7coW2nAULOB=OoOAOGIRuC(eDp%!bL5)|Jq4v z740qOxH34(%)}h?)3+JF{wKS_B3I&NGGCs3=0*HA`LW5{muiicwzf7*CKJt#?Q|Ny zV`$(yGm|%&n|#0@6J~DDFVNc5iM#7YqN5Lzn3zS@K?Qk=d^GARPIq*0?(BK=^(}ne z*~`fAG^VZr`n$*IJ3q$wZ*#OZpT=9}!>--Qq^2k+%qyiJw}@J8HOA&nI$BKBSGO{B zag?i8xqT~~b^UFA4Fi@{;BJAxL8XLNp znVQFJHgosZBGY4LmKK*-6moTV1YfT}o?ItmkHB!e#vriQVKf@q{>E;kQX6V@%@|L9 zOKR@1<#)-F+tZY6$|%^LQnuOm4ro7`-rn43Sg_TD5VWYKl%EUNNKF&7V3+1RlsES5|C z6Wp4+w^H-+@-jTw?eUh!jsKgLg{GxhLVCK1o#IEZdl;v;v@dj ziuy0p+cUV*^M!>)W@m2_x+9!~xFmv}--4ZkE4FrCf|JMiAU%Wh)N~x|9k8^t#BrT7 zLC?u?arWYA-yov)zQ?Ptgo(aXP^c^;B0LTk*8rq89)!FZL0oh!N<|)-2M*!n==2E3 zEwHk(CRo0WPJvZl*GTS1#cU1U&YHC{oSgkhj7>n1nJxNNimFUS$ZH{ZZS+QJB^AAX zikG*(E@u6?;BPy1Iz5MT3fQ?b5?k91tluCbI`UoiC#Q)VITS04P?xJI61uXq5{*hD jW^53={(^elaa5&h4juTAcOqi2wzg-zqx(O2|JV8(TkdM9 From d0dda7769feda81d14efa6266ac5a6c59084b16b Mon Sep 17 00:00:00 2001 From: Luke van der Hoeven Date: Tue, 18 Dec 2012 08:13:23 -0500 Subject: [PATCH 4/6] cleanup content --- _posts/2010-09-01-back-again.html | 3 +-- _posts/2010-09-01-brand-new.html | 2 -- _posts/2010-09-01-collaboration.html | 6 ----- ...10-09-01-cool-ruby-plugin-for-the-day.html | 2 -- _posts/2010-09-01-css-nav.html | 2 -- _posts/2010-09-01-details.html | 7 ----- _posts/2010-09-01-development-to-the-max.html | 1 - .../2010-09-01-dynamic-form-generation.html | 3 +-- _posts/2010-09-01-new-designs.html | 2 -- _posts/2010-09-01-passenger-3-coming.html | 4 --- _posts/2010-09-01-peerdrum-live.html | 2 -- ...0-09-01-podcasting-skip-filter-and-ha.html | 5 +--- _posts/2010-09-01-rails-3-is-coming.html | 1 - _posts/2010-09-01-rails-3-progression.html | 2 -- _posts/2010-09-01-steam-for-mac.html | 1 - ...0-09-01-this-will-only-make-things-wo.html | 1 - _posts/2010-09-08-a-tale-of-two-asyncs.html | 2 -- .../2010-09-10-because-i-told-you-before.html | 1 - _posts/2010-09-23-kthxbye-1-0-0.html | 3 --- _posts/2010-09-25-site.html | 2 -- _posts/2010-09-30-messing-with-the-tubes.html | 3 +-- _posts/2010-09-30-month-in-music.html | 12 --------- .../2010-10-04-ie-gotcha-4-ie7-edition.html | 2 +- .../2010-10-09-new-project-in-the-works.html | 12 --------- _posts/2010-10-25-what-i-have-decided.html | 10 ------- ...1-04-08-a-new-thing-starting-at-sales.html | 4 +-- _posts/2011-06-20-listening-last-week.html | 10 ------- ...1-07-13-mostly-false-ive-been-listeni.html | 4 +-- _posts/2011-07-14-first-jquery-hack.html | 26 +++++++------------ _posts/2011-08-18-designing-for-change.html | 4 +-- 30 files changed, 20 insertions(+), 119 deletions(-) delete mode 100644 _posts/2010-09-30-month-in-music.html delete mode 100644 _posts/2010-10-09-new-project-in-the-works.html delete mode 100644 _posts/2010-10-25-what-i-have-decided.html delete mode 100644 _posts/2011-06-20-listening-last-week.html diff --git a/_posts/2010-09-01-back-again.html b/_posts/2010-09-01-back-again.html index dfd9af71e..cfd3c8a3c 100644 --- a/_posts/2010-09-01-back-again.html +++ b/_posts/2010-09-01-back-again.html @@ -27,5 +27,4 @@

    render "shared/translate"

  end

-

Which is a super fast way of ajaxing the text-area text to an ActionController::Metal action, which faster than thought possible, returns the HTML output version of the Textile markup’d text! Its incredibly fast since I’m only using Metal rather than a full ActionController object. Plus its very simple in what it does, so I’m pretty happy with how it worked out. Check out what it looks like!

-

[[posterous-content:yahilmbvnDwdtAofxsqb]]

+

Which is a super fast way of ajaxing the text-area text to an ActionController::Metal action, which faster than thought possible, returns the HTML output version of the Textile markup’d text! Its incredibly fast since I’m only using Metal rather than a full ActionController object. Plus its very simple in what it does, so I’m pretty happy with how it worked out.

diff --git a/_posts/2010-09-01-brand-new.html b/_posts/2010-09-01-brand-new.html index e28a4dd36..c040f7338 100644 --- a/_posts/2010-09-01-brand-new.html +++ b/_posts/2010-09-01-brand-new.html @@ -11,8 +11,6 @@

LowCountry Community Church

-

[[posterous-content:uucEeeyutwfyFFshgEza]]

-

Still wrestling between domain name providers to transfer the old domain over (from NetworkSolutions to GoDaddy). Just as a technical FYI, NetworkSolutions currently does not allow setting of a single A Record name (say www) to multiple ip addresses (as is needed with Heroku via this article.

Still work to be done in allowing the end admins to be able to customize the copy without my intervention, but that’ll come. This weekend was the deadline to make the switch over from their old static page site to mine. Not too bad….

diff --git a/_posts/2010-09-01-collaboration.html b/_posts/2010-09-01-collaboration.html index 3d8101d33..439830a15 100644 --- a/_posts/2010-09-01-collaboration.html +++ b/_posts/2010-09-01-collaboration.html @@ -7,10 +7,6 @@ posterous_url: http://blog.hungerandthirstdesign.com/post/1047921500/collaboration posterous_slug: post/1047921500/collaboration --- -

[[posterous-content:oAnGiCmbEurGFrJuoIBi]] 

-

&

-

[[posterous-content:FCnppHpryfdqiJtvltbr]]

-

I’m very pleased to announce a new collaboration project between HungerAndthirst Design and TGStudios:

Peerdrum.

@@ -21,6 +17,4 @@

&

The idea behind Peerdrum is that employers should be able to rest easy that their employees are working while on the clock, and employees won’t have to worry about tracking time manually. The app will monitor the employees activities and send screenshots and hours logged back to the project supervisors. Want to jump off the clock, browse Facebook for a while? No problem, just pause the screenshot app on your desktop, spend some you-time, then start it back up again when you’re ready to start working again. Nothing gets logged when you’re off the clock, so all your private data is still untouched and your supervisors won’t have to worry about you playing around on their dime!

-

[[posterous-content:jgbpjnvjmwJxJFzoxuqc]]

-

The initial info can be found at the evangelism site for the project, Peerdrum.com. The project it set to launch sometime in August and I’m really excited to show this baby off. It’s some stuff I’ve never done before and its a pretty slick idea on the whole. Stay tuned…

diff --git a/_posts/2010-09-01-cool-ruby-plugin-for-the-day.html b/_posts/2010-09-01-cool-ruby-plugin-for-the-day.html index 511e1b4f2..416589914 100644 --- a/_posts/2010-09-01-cool-ruby-plugin-for-the-day.html +++ b/_posts/2010-09-01-cool-ruby-plugin-for-the-day.html @@ -10,7 +10,5 @@

Ruby Progressbar: This little bugger is a really simple way to create a textual progress bar for your long-running ruby scripts. I’m working on a script that takes about 30-45 minutes to run. It does some file parsing and then querying of a web service based on the parsed values for reconciliation purposes. The thing is, the web service query takes about 3-5 seconds per query and we’re querying over 7000 records (which means we’d be talking about 7.5 hours to do this if we weren’t also doing some multithreaded magic).

I’m actually planning to run this thing in a few different ways out of personal curiousity. I want to run this in Ruby 1.9.1 with threads and compare that with the time it takes to fork individual processes out for each query. Right now, I’m running it with 15 threads (although I haven’t worked to figure out what an optimum number of threads is).

I also want to see what happens if I load up MacRuby’s implementation of Ruby 1.9.0 and run it under it’s GCD support to see how that compares against the other two. Although I suspect if I’m forking out individual processes, the GCD would be handling those processes anyways? I’m not intimate with the workings of GCD normally or under MacRuby. Time to research!

-

But for now, a screenshot of the progress bar thinger for Ruby scripts. Implementation details are on the script’s website.

-

[[posterous-content:FetjClCyChtFgJIhGDke]]

Update:

Totally forgetting what I learned in parallels class with the idea of trying to use Ruby’s Process library. Process doesn’t have a clean way to aggregate results from processes. It simply spawns off another Ruby instance, runs the given block and exits. So it looks as though if I want to create a concurrent execution version, I will need to either use a whole separate process spawning library… or MacRuby’s GCD. I think we’ll give GCD a chance first.

diff --git a/_posts/2010-09-01-css-nav.html b/_posts/2010-09-01-css-nav.html index f7c3bc3af..c82879b43 100644 --- a/_posts/2010-09-01-css-nav.html +++ b/_posts/2010-09-01-css-nav.html @@ -9,8 +9,6 @@ ---

Okay, so maybe this is news to just me, but I’ve often struggled in the past to get a good looking login/admin control nav that sits neatly up in the top, right-hand corner of the screen without resorting to the unsightly, hard-to-control css float property. Float tends to bounce, jump, wrap, move at bad times… like during client demos. Ehem.

-

[[posterous-content:DDsbprrtgruzBCuyJHig]]

-

So I found a rather elegant solution from a few other sites. Really all it is is setting two divs:

diff --git a/_posts/2010-09-01-details.html b/_posts/2010-09-01-details.html index dac977ff9..9134edc77 100644 --- a/_posts/2010-09-01-details.html +++ b/_posts/2010-09-01-details.html @@ -9,13 +9,6 @@ ---

And the Hastie saga continues… now have Facebook auth and event posting/updating working. Also got the Twitter auth working but haven’t had the time to get the event updating done yet. Should be really simple. Some cool details:

-

-[[posterous-content:meGmgaDjdszrujftsvIi]]

-

And a small little friendly feature to let you know, Hey, you’re socially aware! Good for you! -

-

-[[posterous-content:xrcotgpnIbiecHprCzqf]]

-

Small progresses. Always rewarding. Check it out for yourself at Hastie-app.com

diff --git a/_posts/2010-09-01-development-to-the-max.html b/_posts/2010-09-01-development-to-the-max.html index e313a0718..c8b4bcbf3 100644 --- a/_posts/2010-09-01-development-to-the-max.html +++ b/_posts/2010-09-01-development-to-the-max.html @@ -12,4 +12,3 @@

Tonight it finally happened. So now will you please welcome Hastie! The new way to schedule things. Actually its not new, but its got a lot of potential. I’m always a big fan of single source applications that tie a bunch of resources into a single go-to location. Now I get to try to write one. Lots of ideas in the works for this, most of which will take sometime, or well I suppose they might not since there is already so much functionality out there for some of the features I want to add in. Like Facebook and Twitter for example already have good and diverse APIs that people have written Ruby wrappers for. So I dunno how long it might take to add all the stuff I have planned for it, but its a project. A good one. Check it out, browse around, tell me what you think and if you recognize the design schema from any other projects!

Checkout Hastie!

-

[[posterous-content:flqGdjtlpEyydxpEfAEw]]

diff --git a/_posts/2010-09-01-dynamic-form-generation.html b/_posts/2010-09-01-dynamic-form-generation.html index 5df130481..c7789885b 100644 --- a/_posts/2010-09-01-dynamic-form-generation.html +++ b/_posts/2010-09-01-dynamic-form-generation.html @@ -9,8 +9,7 @@ ---

Interesting app for Hargray Communication this past few weeks. I was asked to design a dynamic survey application that is built, more or less, based on a configuration stored in a database. This is so we can have a front-end application in Oracle that allows, say, non-technical marketing people to build a survey from a form, and then it generates a set of web pages allowing the customer to take the survey.

-

So that’s what I did:

-

[[posterous-content:DAxJvyyvAIrBsboDdAzr]]

+

So that’s what I did

This is a fairly simple app on the whole. Solely purposed to allow customers to take surveys based on a link we send them. So why am I mentioning this?

diff --git a/_posts/2010-09-01-new-designs.html b/_posts/2010-09-01-new-designs.html index 41edc954a..45f9f5793 100644 --- a/_posts/2010-09-01-new-designs.html +++ b/_posts/2010-09-01-new-designs.html @@ -8,11 +8,9 @@ posterous_slug: post/1047054175/new-designs ---

Working on some new site designs, some for the company I work for on a full time basis, others for my church or personal interests. The portfolio is coming along.

-

[[posterous-content:rbteptBEmfGgzcvtrpHB]]

The first one, called very simply: App Monitor, is a very minimal design for a small, yet very useful site I need at work. I wanted a quick, dirty, yet elegant way to view website status, to check if the site was up, and similarly with the actual server machine. So threw together some quick graphics and a light backend to support it (with some slick JS tricks) and here we go. Its not accessible to public use, since it is an internal app, but I was really happy with the design.

-

Uploaded with plasq’s Skitch!

This site has a lot of really interesting ideas working on the back end to do all the site checks and is using some new OSS projects I haven’t had time to play with otherwise. Also of note, this is my first project using Rails 3. I already want to switch all my other projects over, although I’m not sure how practical that is. We shall see. I’ll try to blog about that experience more in the near future.

My other site project is a more professional look, without a lot of flashy elements, but a very clean, information driven design. This one you can access in the public domain, although it is far from complete.

diff --git a/_posts/2010-09-01-passenger-3-coming.html b/_posts/2010-09-01-passenger-3-coming.html index e38330a62..c95b76a2b 100644 --- a/_posts/2010-09-01-passenger-3-coming.html +++ b/_posts/2010-09-01-passenger-3-coming.html @@ -18,8 +18,4 @@

The Road to Passenger 3: Technology Preview 1 – Performance

(continue reading here)

-

Some stats from the post:

- -

[[posterous-content:lnFHqquyBomczaDmFwCh]]

-

Its a fine day to be a rubyist/rails dev.

diff --git a/_posts/2010-09-01-peerdrum-live.html b/_posts/2010-09-01-peerdrum-live.html index b90215fef..fa6ac0fa9 100644 --- a/_posts/2010-09-01-peerdrum-live.html +++ b/_posts/2010-09-01-peerdrum-live.html @@ -10,5 +10,3 @@

Check it out! First major project to hit the big time. More press material coming soon!

peerdrum.com

-

[[posterous-content:bqrzlppghsfbnJaetket]]

-

[[posterous-content:eqpuormiicHbItHGnqdq]]

diff --git a/_posts/2010-09-01-podcasting-skip-filter-and-ha.html b/_posts/2010-09-01-podcasting-skip-filter-and-ha.html index 4ea0c84a8..bc1e98d44 100644 --- a/_posts/2010-09-01-podcasting-skip-filter-and-ha.html +++ b/_posts/2010-09-01-podcasting-skip-filter-and-ha.html @@ -11,8 +11,6 @@

Well an interesting weekend for sure. Hastie is still coming along. Did some really stupid stuff while working on it late one night whilst trying to integrate facebook connectivity. mmangino’s amazing Facebooker plugin for Rails works really well if you follow the instructions. Warning: DO NOT ADD “require ‘facebooker’” anywhere in the code. It creates traceless stack crashes which are never fun to debug at 2 in the morning. I ended up resetting my database in an attempt to solve it and deleted some important static data I was holding there. Still kicking myself over that one. In any case, Hastie is coming right along. I should be posting a major update later this week.

-

[[posterous-content:qvkknybAgwjmaIoDcICG]]

-

skip_filter

But this isn’t a post about Hastie. Wanted to post just some interesting things I learn about Rails as I go along on some various projects. Hargray Communication has a massive Oracle database that I have to write my apps against. Oracle is not particularly “Rails-friendly” although its more the fact that our db schema/naming conventions are not Rails approved. So it always takes a lot of wrangling. Why is this a problem? Well, because our user tables are set in stone, most of the nice authentication plugins out there (authlogic, restful-authentication) don’t work (at least by default) with our db schemas. Therefore, I end up writing a lot of custom authenticators. I suppose the cool thing to do would be to fork and mod something like authlogic to work with our databases, which thinking about that now, wouldn’t be too hard since it would mostly involve removing functionality. But anyways, I wrote a very simple auth system using almost exclusively filters. Take a look!

@@ -36,8 +34,7 @@

skip_filter

Podcasting

Added podcast hosting/upload ability to theLowCountry Community Church website yesterday. Took a few hours, but it really wasn’t super difficult.

-

[[posterous-content:bDBGxbjkylFkfzCbfEap]]

-

The image above is a placeholder for what a podcast entry on the page looks like. You can actually subscribe to the feed in iTunes under the “Advance > Subscribe to Podcast” option. I expect it will take some time to show up in the iTunes store podcast listing (submitted yesterday). I didn’t realize until I started researching this project, but a podcast feed is just an XML rss feed. Ours is found here. It just involves a mess of description tags. Our files are actually being hosted on Amazon S3 (for less than 5 cents a month) and I’m pretty stoked with how easy the whole thing was. A lot of the more dynamic content is stored there simply because its so easy to upload the images/audio there. The aws-s3 gem is fantastic for accomplishing the uploading/linking files. A bit of a gotcha in the whole process is ensuring that the files are accessible to the general public with no authentication required. otherwise iTunes complains. Its pretty simple though. My upload and link generating methods look like this:

+

You can actually subscribe to the feed in iTunes under the “Advance > Subscribe to Podcast” option. I expect it will take some time to show up in the iTunes store podcast listing (submitted yesterday). I didn’t realize until I started researching this project, but a podcast feed is just an XML rss feed. Ours is found here. It just involves a mess of description tags. Our files are actually being hosted on Amazon S3 (for less than 5 cents a month) and I’m pretty stoked with how easy the whole thing was. A lot of the more dynamic content is stored there simply because its so easy to upload the images/audio there. The aws-s3 gem is fantastic for accomplishing the uploading/linking files. A bit of a gotcha in the whole process is ensuring that the files are accessible to the general public with no authentication required. otherwise iTunes complains. Its pretty simple though. My upload and link generating methods look like this:

  def get_podcast_image(name)

diff --git a/_posts/2010-09-01-rails-3-is-coming.html b/_posts/2010-09-01-rails-3-is-coming.html index 165f307d3..a74af338d 100644 --- a/_posts/2010-09-01-rails-3-is-coming.html +++ b/_posts/2010-09-01-rails-3-is-coming.html @@ -10,4 +10,3 @@

Beta 4 is out!

Also a big step for Bundler which just hit 1.0.0.beta1

So much coming soon!

-

[[posterous-content:wraaolohjjjvFgiCkffg]]

diff --git a/_posts/2010-09-01-rails-3-progression.html b/_posts/2010-09-01-rails-3-progression.html index d2163234e..c76de2482 100644 --- a/_posts/2010-09-01-rails-3-progression.html +++ b/_posts/2010-09-01-rails-3-progression.html @@ -7,8 +7,6 @@ posterous_url: http://blog.hungerandthirstdesign.com/post/1047827560/rails-3-progression posterous_slug: post/1047827560/rails-3-progression --- -

[[posterous-content:DhJBjbofvmtBtakfndrr]]

-

Check it out!! Converted this site to Rails 3 beta 3. Very cool. Way to go Rails team. Keep up the work.

UPDATE: Apparently Heroku does not actually allow beta 3 yet. I think it require Ruby 1.9.2 which is still in pre-release state. So close…

diff --git a/_posts/2010-09-01-steam-for-mac.html b/_posts/2010-09-01-steam-for-mac.html index 42c350dfe..e59c56ad4 100644 --- a/_posts/2010-09-01-steam-for-mac.html +++ b/_posts/2010-09-01-steam-for-mac.html @@ -12,7 +12,6 @@

Its here.

-[[posterous-content:zBhbDEEIEGCzjbGDzHqH]]

First impressions are just that the interface is really well done. Very clean, intuitive, and it works! Doesn’t look like the store is up on the client yet, but it exists on Steam’s site. Mostly some slick indie games and Portals!! Now we just wait for Portals 2…

diff --git a/_posts/2010-09-01-this-will-only-make-things-wo.html b/_posts/2010-09-01-this-will-only-make-things-wo.html index 194c64832..836b183dd 100644 --- a/_posts/2010-09-01-this-will-only-make-things-wo.html +++ b/_posts/2010-09-01-this-will-only-make-things-wo.html @@ -7,7 +7,6 @@ posterous_url: http://blog.hungerandthirstdesign.com/post/1047935800/this-will-only-make-things-worse posterous_slug: post/1047935800/this-will-only-make-things-worse --- -

[[posterous-content:roGtybwalHcGBDmDamkz]]

I already entered “:wq” in way too many other editors. This certainly isn’t going to help.

Starting up with vim as my primary editor. Going to take some time to get used to, but I really like it so far. I always wanted to use keyboard only navigation…

diff --git a/_posts/2010-09-08-a-tale-of-two-asyncs.html b/_posts/2010-09-08-a-tale-of-two-asyncs.html index 74e187c7d..e3845ca4b 100644 --- a/_posts/2010-09-08-a-tale-of-two-asyncs.html +++ b/_posts/2010-09-08-a-tale-of-two-asyncs.html @@ -13,8 +13,6 @@

There are a number of projects I can think of where a job that takes longer than a user ought to be waiting for a result (due to server timeout length or out of simple app responsiveness courtesy to the user). At the same time I need the result of the operation to be returned to the user.

Here’s a real-world example. I work with a set of legacy Oracle databases that stores much of our business logic as PLSQL procedures. Yes, this is not “The Rails Way™” but it’s the only way for the company I work for right now. Many of the procedures that I run as part of several of the applications I support can take on average one minute or more with a standard deviation of almost 2 minutes (with a forced timeout of 5 minutes). That’s kinda a long time to sit and wait on a web app regardless of what web app we’re talking about.

-

[[posterous-content:olburAEvomACaqifsxgl]]

-

We don’t really want users sitting waiting for up to 5 minutes (at which it forces failure) unable to do anything or (even worse) hitting refresh or the action again. Especially bad when this can mean the HTTP server is getting backed up as more and more people run these long running processes.

Moreover, the users need to get response from the completed job before moving on. Most job processors (DJ, Resque) are setup for running jobs that do not require the result to be returned to the web app (think mass-mailers, queue population, image resizing). They just run and the output goes to a database, an inbox or a file server.

diff --git a/_posts/2010-09-10-because-i-told-you-before.html b/_posts/2010-09-10-because-i-told-you-before.html index a39fa7235..b9802afc0 100644 --- a/_posts/2010-09-10-because-i-told-you-before.html +++ b/_posts/2010-09-10-because-i-told-you-before.html @@ -8,7 +8,6 @@ posterous_slug: post/1095348551/because-i-told-you-before ---

No, you can’t do that.

-

[[posterous-content:xnmGwuhEHtfruyjyDcxa]]

So I’m gonna do a bit of backtracking on what I told you before. After a good couple of days of futzing about with fibers, EventMachine and Processes and have made some interesting discoveries about how to go about things when working out concurrency.

Fibers are awesome, but they’re a bit tricky when it comes to using at a higher level around other libraries. Why? Because there is no good way to preempt fibers when it hits a blocking process… such as a long DB call, which is the whole point. In order to get that level of preemption, I’d have to make all the supporting libraries fiber aware as well. That’s a lot of work for this project. It would involve, at least in this instance, modifying Raimonds Simanovskis PLSQL adapter and probably the oracle_enhanced gem as well and that’s way beyond the scope of this project. Probably worth doing, just not what I’m working on at the moment.

So that leaves EM and Processes. I worked with EM for a bit and discovered its a bit tricky to use within an infinite loop. It works, just not as well as I’d like and without the level of concurrency or controllability I would like. Deferrables are nice, but it’s hard to spawn off multiple defers for each of the blocking processes that might be run in parallel. 

diff --git a/_posts/2010-09-23-kthxbye-1-0-0.html b/_posts/2010-09-23-kthxbye-1-0-0.html index 8466e6d42..44f63dd5b 100644 --- a/_posts/2010-09-23-kthxbye-1-0-0.html +++ b/_posts/2010-09-23-kthxbye-1-0-0.html @@ -9,11 +9,8 @@ ---

Hey all. quite pleased to introduce the “final” version of my latest project, Kthxbye. As I wrote about in the last few blog posts, this gem allows you farm out long running processes that you need a result from to a separate processing engine so that you don’t have to tie up your web-server with long running processes.

This doubtlessly will have many revisions in the future but it took me so long to finalize the initial API that I wanted to solidify something for an initial release. This library is fully tested for Ruby 1.8.7 and 1.9.2. 

-

[[posterous-content:dgHmfgBHgoFreGqIlEDz]]

It has a pretty slick web front-end too. Lots of useful information to display as well as a live polling feature which refreshes the data on the screen every 2 seconds.

-

[[posterous-content:AwddHqhowJzfxtbzvAsc]]

It also gives you good indication as to what the status of the job is, when it’s completed, whether or not it succeeded or failed. This is all available through the API, and therefore, to your app as well.

-

[[posterous-content:sGofbgqHpChsgyDaacJu]]

The interface is built on Sinatra and should be pretty easy to integrate into your Rails app stack (made especially easy in Rails 3).

So how about some comments on the project? This is my first time doing any kind of major unit testing. Definitely will be doing more of this in the future. Incredibly helpful to know when I broke something earlier, caught so many more exceptions and edge cases I might have otherwise missed. Obviously that’s the point of unit testing, but it was cool to see just how helpful it was long term.

Where to get this sugary goodness?

diff --git a/_posts/2010-09-25-site.html b/_posts/2010-09-25-site.html index 0a2235d6e..be216b812 100644 --- a/_posts/2010-09-25-site.html +++ b/_posts/2010-09-25-site.html @@ -8,7 +8,6 @@ posterous_slug: post/1183754580/site ---

Finally made the first of many updates to my site. Color scheme change is next. Or maybe the reskinning of this blog?

-

[[posterous-content:joHjmiixqBsCjuFAywmu]]

In any case, I’m now with webbynode. I figured if anything, my site ought to be the first one to make a switch as I’m getting a little discontent with Heroku’s availability some days. Moreover, their charges for additional services are getting outrageous. 

It’s not incredibly difficult to setup. Definitely some changes over Heroku, who I must say still have one of the best deploy workflows out there. But what is incredibly valuable is I now have root access to the server, which means I can do just about anything I want to it. I can kill it in cold blood if I want. Big. Fat. “shutdown -t now”, But lets not get hasty…

The big benefit is I can now run Node.JS, Redis, MongoDB, custom ruby tasks, cron jobs… anything I want… for free. Heroku is charging an arm and a leg for that privilege. 

@@ -19,4 +18,3 @@

New experiment too: Trying on Nginx for the first time. Running passenger on top of that for now. Seems like a really clean way to do web server configuration. I like. Seems fast enough. Consumes very little memory which is good on my lower-end node.

So all in all, I like Webbynode so far. I’ll be interested to start integrating more stuff into it like Redis or some other db backend. I guess in a sentence to sum it up: root access is awesome.

-

[[posterous-content:CFejrabuxquhwCGliICa]]

diff --git a/_posts/2010-09-30-messing-with-the-tubes.html b/_posts/2010-09-30-messing-with-the-tubes.html index e99976ff8..8e0eaaeb6 100644 --- a/_posts/2010-09-30-messing-with-the-tubes.html +++ b/_posts/2010-09-30-messing-with-the-tubes.html @@ -11,8 +11,7 @@

However, this trip gets pretty crazy… and doesn’t work quite right yet, at least not in… dun-du-dun… Internet Explorer. But that’s another problem to solve as we go along (http://stackoverflow.com/questions/3825271/socket-io-node-js-rails-3-ie).

So there is now a second optional module to add to Kthxbye when using it behind a Rails application. I’ve designed a slick little notification widget that will be updated in realtime when the job gets picked up to be worked, once it completes or fails. All you’ll need to do in the end is include a CSS file, a javascript file and a

<div id='kthxbye'></div>

-

tag to your html or other markup and you’ll get this (or well, the first of the four blocks).

-

 [[posterous-content:gjerkuFGnkteaEdonGEb]]

+

tag to your html or other markup.

Then once you startup a worker, it’ll change the colors of the lights based on the status of the job.

How does this work? Its a three layered system. Kthxbye uses Redis’ PUBSUB protocol to publish a message to the “job.*” channel, where “*” will be success|failed|active. There is then a Node.js server sitting in-between the Rails app and Redis that subscribes to the “job.*” channel waiting for messages to come down. The message currently only contains a single number with the original job id that it was queued under, this way when the clients receive the message, they can filter for what message they want on the client-side rather than opening up potentially infinite number of channels for each job, which as it appears from the Redis docs would ultimately slow down publishing to the messaging queues as per this specification: 

diff --git a/_posts/2010-09-30-month-in-music.html b/_posts/2010-09-30-month-in-music.html deleted file mode 100644 index 536873810..000000000 --- a/_posts/2010-09-30-month-in-music.html +++ /dev/null @@ -1,12 +0,0 @@ ---- -layout: post -title: Month in music -published: true -date: 2010-09-30 -categories: [] -posterous_url: http://blog.hungerandthirstdesign.com/post/1216931726/month-in-music -posterous_slug: post/1216931726/month-in-music ---- -

Been a long time:

-

[[posterous-content:jlcJDImjbjuHuBfDAzeB]]

-

[[posterous-content:fGtdeBCqsvwyorxAnrqv]]

diff --git a/_posts/2010-10-04-ie-gotcha-4-ie7-edition.html b/_posts/2010-10-04-ie-gotcha-4-ie7-edition.html index 41341e076..612bf6722 100644 --- a/_posts/2010-10-04-ie-gotcha-4-ie7-edition.html +++ b/_posts/2010-10-04-ie-gotcha-4-ie7-edition.html @@ -7,7 +7,7 @@ posterous_url: http://blog.hungerandthirstdesign.com/post/1243153618/ie-gotcha-4-ie7-edition posterous_slug: post/1243153618/ie-gotcha-4-ie7-edition --- -

[[posterous-content:GftujlwFpHolACdCAumG]]In an ongoing attempt to alert the general public to all the downfalls you suffer when you use IE I continue the series: IE Gotchas. I haven’t hit any major ones in a while, but this is a weird one.

+

In an ongoing attempt to alert the general public to all the downfalls you suffer when you use IE I continue the series: IE Gotchas. I haven’t hit any major ones in a while, but this is a weird one.

IE7 (and maybe before?) doesn’t give you the “JSON” object with which to manipulate json objects. Whaaat?

Simple solution is to require a conditional comment to include this file somewhere. Then you’re fine.

<!-- [if lte IE 7]> <script src="http://webscan.googlecode.com/svn-history/r80/trunk/ui/js/third-party/json2.min.js" type="text/javascript"></script> <![endif] -->

diff --git a/_posts/2010-10-09-new-project-in-the-works.html b/_posts/2010-10-09-new-project-in-the-works.html deleted file mode 100644 index df75a3b05..000000000 --- a/_posts/2010-10-09-new-project-in-the-works.html +++ /dev/null @@ -1,12 +0,0 @@ ---- -layout: post -title: New Project in the Works -published: true -date: 2010-10-09 -categories: [] -posterous_url: http://blog.hungerandthirstdesign.com/post/1274087562/new-project-in-the-works -posterous_slug: post/1274087562/new-project-in-the-works ---- -

More design focused this time. Sneak peak:

- -

[[posterous-content:iDFApilBvGqzFpmczlxs]]

diff --git a/_posts/2010-10-25-what-i-have-decided.html b/_posts/2010-10-25-what-i-have-decided.html deleted file mode 100644 index 57faeb556..000000000 --- a/_posts/2010-10-25-what-i-have-decided.html +++ /dev/null @@ -1,10 +0,0 @@ ---- -layout: post -title: Untitled -published: true -date: 2010-10-25 -categories: [] -posterous_url: http://blog.hungerandthirstdesign.com/post/1398983003/what-i-have-decided -posterous_slug: post/1398983003/what-i-have-decided ---- -

[[posterous-content:xDfvszmIrCwpizGgeIJw]]

What I have decided.

diff --git a/_posts/2011-04-08-a-new-thing-starting-at-sales.html b/_posts/2011-04-08-a-new-thing-starting-at-sales.html index 674f19825..29c1d6a58 100644 --- a/_posts/2011-04-08-a-new-thing-starting-at-sales.html +++ b/_posts/2011-04-08-a-new-thing-starting-at-sales.html @@ -1,10 +1,10 @@ --- layout: post -title: Untitled +title: Stevie published: true date: 2011-04-08 categories: [] posterous_url: http://blog.hungerandthirstdesign.com/post/4446525388/a-new-thing-starting-at-salescrunch-we-have-a posterous_slug: post/4446525388/a-new-thing-starting-at-salescrunch-we-have-a --- -

[[posterous-content:zIfAGwGEtCmBfagxyynv]]

A new thing starting at SalesCrunch. We have a mascot, and his name is Steve.

+

A new thing starting at SalesCrunch. We have a mascot, and his name is Steve.

diff --git a/_posts/2011-06-20-listening-last-week.html b/_posts/2011-06-20-listening-last-week.html deleted file mode 100644 index f6d86c8e8..000000000 --- a/_posts/2011-06-20-listening-last-week.html +++ /dev/null @@ -1,10 +0,0 @@ ---- -layout: post -title: Untitled -published: true -date: 2011-06-20 -categories: [] -posterous_url: http://blog.hungerandthirstdesign.com/post/6724483902/listening-last-week -posterous_slug: post/6724483902/listening-last-week ---- -

[[posterous-content:hmnyahuwdIEbetrrwCbb]]

Listening last week

diff --git a/_posts/2011-07-13-mostly-false-ive-been-listeni.html b/_posts/2011-07-13-mostly-false-ive-been-listeni.html index ffdf0f4b8..ec0091ea9 100644 --- a/_posts/2011-07-13-mostly-false-ive-been-listeni.html +++ b/_posts/2011-07-13-mostly-false-ive-been-listeni.html @@ -1,10 +1,10 @@ --- layout: post -title: Untitled +title: Music published: true date: 2011-07-13 categories: [] posterous_url: http://blog.hungerandthirstdesign.com/post/7559745380/mostly-false-ive-been-listening-to-this-for-the posterous_slug: post/7559745380/mostly-false-ive-been-listening-to-this-for-the --- -

[[posterous-content:CwEHJAGkIeadjigxjChh]]

Mostly false. I’ve been listening to this for the past three days straight. Radiohead is soooo good.

+

I’ve been listening to this for the past three days straight. Radiohead is soooo good.

diff --git a/_posts/2011-07-14-first-jquery-hack.html b/_posts/2011-07-14-first-jquery-hack.html index de87d8218..e0490e0f1 100644 --- a/_posts/2011-07-14-first-jquery-hack.html +++ b/_posts/2011-07-14-first-jquery-hack.html @@ -13,25 +13,17 @@

As referenced often before in previous blog posts, SalesCrunch is using a wide diversity of technologies to move people around, especially during a live meeting. It’s heavily centered around Node.js and Redis to handle async messaging between server and client. Because of this, there are a lot of failure points for us, and without having access to client’s browser info, it can be hard to debug beyond the stack traces we get. So I was tasked with setting up a diagnostics page, where you could check all the variables and connections that our servers make to your browser.

-

It looks something like this:

- -

[[posterous-content:qGwIbbbczJatqpsGjdId]]

-

We cover a lot of the items we require to use our software: JS, Java, Flash, Socket connectivity, and so on. It was fairly easy to tackled most of them.

Flash: Simply checked if there is a flash plugin installed via the browser’s navigator.plugin object. JS is simply populating the field with enabled… via javascript. If it’s disabled, it uses <noscript> tags to populate. Java is a mini-applet that simply provides information on the JVM. Browser and OS info can be determined from the HTTP user-agent request header. We can query our socket (Socket.IO) to see if it’s connected or not.

The last one, bandwidth test, was the most interesting, mostly because it involves timing and math skills, both of which I generally fail at. So I inquired of google for some ideas. The general idea is fairly simple: Download something of a known file size and time it. Bandwidth, is after all Mbits per second (not to be confused, as I did earlier with MBytes… ehem). This should be easy then.

-

The trick is: It’s not entirely easy. Everything you hear about trying to calculate bandwidth by this method is working around server response delays. You can see it pretty quick in Chrome when running locally:

- -

[[posterous-content:voBrdwHqwznsJsAzDwzs]]

+

The trick is: It’s not entirely easy. Everything you hear about trying to calculate bandwidth by this method is working around server response delays.

-

Now, the upside is that you don’t have to worry about this as much, so long as your file is big enough to make delay negligible and not count for much against the total download time, like so:

+

Now, the upside is that you don’t have to worry about this as much, so long as your file is big enough to make delay negligible and not count for much against the total download time

-

[[posterous-content:HvfdsGjdniDfvddthslt]]

- -

As you can see, the waiting time is not a big percentage of the overall download time. However, I wanted to do better, partly out of curiosity, partly out of pursuing excellence. Since I’m using jQuery, I figured it shouldn’t be too hard to track down a way to detect the first response back from the server. More importantly, I want to know when the actual file data (in this case, a screenshot from Minecraft) is being returned, minus the HTTP header info. Precious bytes…

+

The waiting time is not a big percentage of the overall download time. However, I wanted to do better, partly out of curiosity, partly out of pursuing excellence. Since I’m using jQuery, I figured it shouldn’t be too hard to track down a way to detect the first response back from the server. More importantly, I want to know when the actual file data (in this case, a screenshot from Minecraft) is being returned, minus the HTTP header info. Precious bytes…

Digging into the jQuery code, they do, in fact track a readyState, very similar to the one specified by XMLHttpRequest. This has a number of important states that we care about:

@@ -39,19 +31,19 @@

The XMLHttpRequest object can be in several states. The readyState attribute must return the current state, which must be one of the following values:

- +

UNSENT (numeric value 0) The object has been constructed.

- +

OPENED (numeric value 1) The open() method has been successfully invoked. During this state request headers > can be set using setRequestHeader() and the request can be made using the send() method.

- +

HEADERS_RECEIVED (numeric value 2) All redirects (if any) have been followed and all HTTP headers of the final response have been received. Several response members of the object are now available.

- -

LOADING (numeric value 3) + +

LOADING (numeric value 3) The response entity body is being received.

- +

DONE (numeric value 4) The data transfer has been completed or something went wrong during the transfer (e.g. infinite redirects).

diff --git a/_posts/2011-08-18-designing-for-change.html b/_posts/2011-08-18-designing-for-change.html index 945c8ae04..67a52cb3a 100644 --- a/_posts/2011-08-18-designing-for-change.html +++ b/_posts/2011-08-18-designing-for-change.html @@ -13,7 +13,7 @@

A Real Life Example

Our product, as I have noted before, is a presentation and tracking tool, purposed to allow users to present slide decks, normally as part of a sales presentation and provide the presenter with tracking metrics both in live (presenter-driven) and emailed (viewer-driven) meetings or presentations. As part of this, we allow the users to upload and/or select decks for use in presentations or to send to users. We have an interface for doing this that used to exist in two parts: One for uploading, naming and organizing decks, the other for selecting decks for use. As part of one level of redesign, these two interfaces were forced together so that, during a live meeting, a presenter could both upload and then use a deck during a live meeting. The result looked something like the following two screens.

-

[[posterous-content:GdzJgHItbfFEhCniAsbq]] +

Image 1: The Original Selection Screen

(Please ignore the design flaws, some of these shots were of prototype interfaces.)

@@ -22,7 +22,7 @@

A Real Life Example

The problem shows up when our design/product team decides the interface is not very straightforward, and I agree, mainly in the separation of the select/upload flow. Our new design separates out the two into separate tabs as seen here:

-

[[posterous-content:yquovwDcJeokGfyiJhGF]] +

Image 2: The New Selection Screen

This makes it easy to see here are my decks and if I need more, I go to the upload tab, there are no major redirections or interface changes.

From 603bfd992ee2917aa63e13a13c1c77d61d912e72 Mon Sep 17 00:00:00 2001 From: Luke van der Hoeven Date: Tue, 18 Dec 2012 08:22:02 -0500 Subject: [PATCH 5/6] add draft post --- _posts/2012-12-18-new-job-3.markdown | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) create mode 100644 _posts/2012-12-18-new-job-3.markdown diff --git a/_posts/2012-12-18-new-job-3.markdown b/_posts/2012-12-18-new-job-3.markdown new file mode 100644 index 000000000..ea2486434 --- /dev/null +++ b/_posts/2012-12-18-new-job-3.markdown @@ -0,0 +1,22 @@ +--- +layout: post +title: ! 'Job.create(company: "BFG Communications")' +published: false +date: 2012-12-18 +categories: [] +--- +Two years later, and it's on to the next one. I started a job with a new kind of company doing a new kind of job. BFG is a creative agency, dealing with a lot of really cool companies and products. They started in the traditional print and personal ad space and have, in the past 6 years or so, moved into the digital space via interactive design, primarily on the web. It's an odd position for me for a number of reasons, the primary one being that their main technology stack is all Microsoft based, in the form of .NET / PHP. ***Horrors*** + + +But seriously… why those two!?! + + +But I kid. There are a lot of really smart people doing some amazing work in this space. What's even cooler is I've gotten to write my own job description for what I will be doing. My official title is "Software Engineer" which to most people means "developer". But in this case, I will end up wearing a wide variety of hats. + + +My primary roles will be to provide the following: + + + +* Provide visibility into the company's process via tools to track work items and create data-driven stories about how the company's pipeline actually works. There is currently no way to do this, so I get to build awesome tools! +* Create developer happiness through teaching better development patterns and design and building reusable components while cutting out as much "reinventing the wheel" as possible. From 639cebbf5c6355b0f579ceb6d181fbe4b68350ff Mon Sep 17 00:00:00 2001 From: Luke van der Hoeven Date: Tue, 22 Jan 2013 09:54:41 -0500 Subject: [PATCH 6/6] update and release post --- _posts/2012-12-18-new-job-3.markdown | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) diff --git a/_posts/2012-12-18-new-job-3.markdown b/_posts/2012-12-18-new-job-3.markdown index ea2486434..98bb7fdbe 100644 --- a/_posts/2012-12-18-new-job-3.markdown +++ b/_posts/2012-12-18-new-job-3.markdown @@ -1,22 +1,24 @@ --- layout: post title: ! 'Job.create(company: "BFG Communications")' -published: false +published: true date: 2012-12-18 categories: [] --- Two years later, and it's on to the next one. I started a job with a new kind of company doing a new kind of job. BFG is a creative agency, dealing with a lot of really cool companies and products. They started in the traditional print and personal ad space and have, in the past 6 years or so, moved into the digital space via interactive design, primarily on the web. It's an odd position for me for a number of reasons, the primary one being that their main technology stack is all Microsoft based, in the form of .NET / PHP. ***Horrors*** - But seriously… why those two!?! - But I kid. There are a lot of really smart people doing some amazing work in this space. What's even cooler is I've gotten to write my own job description for what I will be doing. My official title is "Software Engineer" which to most people means "developer". But in this case, I will end up wearing a wide variety of hats. - My primary roles will be to provide the following: +* Provide visibility into the company's process via tools to track work items and create data-driven stories about how the company's + pipeline actually works. There is currently no process or tool for this, so I get to build awesome tools! +* Create developer happiness through teaching better development patterns and design and building reusable components while + cutting out as much "reinventing the wheel" as possible. +* Setup infrastructure, including CI/CD tools and better build tools to integrate things like Less and Coffeescript into the general + workflow. - -* Provide visibility into the company's process via tools to track work items and create data-driven stories about how the company's pipeline actually works. There is currently no way to do this, so I get to build awesome tools! -* Create developer happiness through teaching better development patterns and design and building reusable components while cutting out as much "reinventing the wheel" as possible. +There are many more things I get to do here, but what all those are remains to be seen. AS I said, it's a very different kind +of position that will give me a lot of flexibility and opportunity to grow as an engineer.