On 07/28/2011 11:49 AM, Rasmus Lerdorf wrote:
On 07/28/2011 12:42 AM, Dmitry Stogov wrote:
On 07/28/2011 11:23 AM, Rasmus Lerdorf wrote:
On 07/28/2011 12:20 AM, Dmitry Stogov wrote:
Hi Rasmus,
Wrong decision.
You make PHP slower just because can't solve an APC issue. :(
I believe the problem may be solved in APC itself.
Please, revert the patch.
PHP is even slower if APC doesn't work with it. I don't see how this can
be solved entirely on the APC side. I'm open to suggestions.
APC is not a single opcode cache, so you are slowing down the others.
I probably can find a solution to fix APC, if I get permission to do it.
But, anyway, please revert the patch. I spent weeks developing the
interned strings solution and now you are slowing it down before
listening to my opinion.
You don't need permission to contribute to APC. Go ahead. We've spent a
long time on APC as well, and the interned strings broke it. Right now
we have a stable state where both PHP and APC work. The patch didn't
break other opcode caches it only slowed them down a little bit.
There
are a large number of sites out there that rely extensively on APC and
they haven't been able to test 5.4 in their environments because of this
problem.
We can revert for now, but we have another alpha release scheduled for
next week and I want all the folks who have been blocked by APC being
broken to be able to participate in this release so we need a solution
we can all live with by then.
Yeah, revert please. I'll take a look into APC after that.
Thanks. Dmitry.
-Rasmus