revert: restore line printing in tracer

[guillett]

Fixes #1310
Depends on #1321

Technical changes

• Fix computation_log.

[bonjourmauko]

Hi! Was this a bug introduced recently? I'd love to add a failing test for this. If you could point me out at how to reproduce the error and the expected behaviour, I could help with the test.

[bonjourmauko]

Ok I understood now: printing/logging has nothing to do in the computation log, which is just a log accumulator —like a writer monad transformer, sort of.

In test_runner.py, you could either:

1. Get the log buffer and log it to stdout there.
2. Pass down a function to the tracer with how you want to print/log the log buffer.

Something like this should give you what you want (just an idea):

# task_runner.py
@staticmethod
def print_computation_log(tracer, aggregate, max_depth) -> None:
    buffer = tracer.computation_log.lines(aggregate, max_depth)
    for line in buffer:
        print(line)

Also, using print is making the linter fail in CI, you have to use a logger for that:
https://docs.python.org/3/library/logging.html

Hope it helps :)

BTW, I have no idea how to test test_runner.py, so I won't be picky with that.

(Also, if this wasn't working before, I'd do a minor bump and label it as a new feature. It it was working before, then you can just safely ignore this comment and keep it as a patch.)

[guillett]

Thanks @bonjourmauko , I don't have much time to follow up. To me, this is a regression and I need this fix to work locally.

[bonjourmauko]

@guillett If you give me the command you normally use I can propose you a stacked PR with the proposed fix.

Concerning print, even if I accept it, you won't be able to merge this PR because of the linter.

[bonjourmauko]

Oh nevermind, I see the regression was introduced here 2f8072f#diff-fc2477fbba8037857fb9287895920ee5a1015c5eda4edb50832af14232fdd10b

[benjello]

@guillett : I modified the changlog so the linter passes but we still have a problem with a conda build.
@benoit-cty: didn't we got this error already ? Any hint ?

[benoit-cty]

@guillett : I modified the changlog so the linter passes but we still have a problem with a conda build. @benoit-cty: didn't we got this error already ? Any hint ?

Well, the error in Conda complains about Numpy :

If using `np.array(obj, copy=False)` replace it with `np.asarray(obj)` to allow a copy when needed (no behavior change in NumPy 1.x).

Because despite the np1.6.24 in the name of the test, it is a Numpy 2 version that has been installed:

+ numpy 2.2.3 py310hefbff90_0 conda-forge 8MB

[bonjourmauko]

There is definitely a problem with the conda's config.

[bonjourmauko]

poke @sylvainipp it looks like the problem you were having could come from inadvertently using a numpy > 2.2 version, that could be unrelated to your proposed changes in #1319