Control caching based on response

[volodymyrss]

I might be missing entierly the intent of the developers, and possibly there is a much better solution to my problem (#91).
This is just what I implemented which is sufficient for my needs, I wonder if it makes any sense.

[coveralls]

Coverage increased (+0.02%) to 79.94% when pulling 50520b9 on volodymyrss:control_with_response into 6140d2c on sh4nks:master.

[coveralls]

Coverage increased (+0.02%) to 79.94% when pulling 50520b9 on volodymyrss:control_with_response into 6140d2c on sh4nks:master.

[coveralls]

Coverage increased (+0.02%) to 79.94% when pulling 50520b9 on volodymyrss:control_with_response into 6140d2c on sh4nks:master.

[coveralls]

Coverage increased (+0.02%) to 79.94% when pulling a5e54f6 on volodymyrss:control_with_response into 6140d2c on sh4nks:master.

[sh4nks]

This looks good to me from a first look! I'll take a closer look when I am home from work.

[volodymyrss]

thanks for the suggestion!
I see you do not use print even in the test, I should have followed that. Could I ask why, for educational purposes?

[sh4nks]

Wait, you were able to commit my suggested changes directly from github? 😄

Generally speaking, instead of using print statements, the information should be logged using a dedicated logger. The logger can then be configured to only print certain log levels to the console, or log everything to a file.

In our case, by reading the test cases you have written one can easily guess what the result is -- and in case the tests fail, pytest will show us all the needed information anyway.

Hope that clears things up!

[volodymyrss]

Ok, I see, thanks!
I would usually use logger except for the tests, but then no reason to treat tests as second-class code.

Yeah, the technology of github is growing beyond our imagination. There was just a button to accept the suggestion, first time I see such a feature.