Use element reference `key` to pipe output of a flow as an input to another flow

[theweaklink]

Change request type

• New feature or Improvement?
  • New feature
  • Improvement
• Is it OS-specific?
  • Not particularly
  • Linux
  • Windows
• Who is it for?
  • Expert
  • Tech-Savvy
  • Developer

Details

• Description:
  Currently you can pipe a file output from a flow as input to another flow by using its path as it is the case for the DeepLearning example:

# flow: train.py
model.save(file_output('model', 'saved_model', tags=['Keras', 'Binary', 'Model']))

# flow: predict.py
model = keras.models.load_model(Project().get_output_path('saved_model'))

This could be error-prone: if we change the path to saved_model in the 1st flow, then the 2nd flow will fail as the output/input is no longer in sync. A way to avoid this potential problem would be to use the output by its key.

• Sample code:

# flow: train.py
model.save(file_output('model', 'saved_model', tags=['Keras', 'Binary', 'Model']))

# flow: predict.py
model = keras.models.load_model(Project().get_output('model'))

This way, even if we change saved_model in the 1st step, the predict.py will still work.

• Desired screenshot/output:
  Project.get_output() returns the value of the output element.