Fix callback default (horror bug!)

[awaelchli]

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure to update the docs?
• Did you write any new necessary tests?
• If you made a notable change (that affects users), did you update the CHANGELOG?

What does this PR do?

Fixes a horror bug that took me sooo many hours to find when implementing #1450.
An argument in a function or method should NEVER be initialized to an empty list! Here is an example of what happens when you do so:

from pytorch_lightning import Trainer
from pytorch_lightning.callbacks import ModelCheckpoint


a = Trainer()
a.callbacks.append(ModelCheckpoint('./a'))

print(len(a.callbacks))

b = Trainer()
b.callbacks.append(ModelCheckpoint('./b'))

print(len(b.callbacks))

Output (on master):

1
2

Expected:

1
1

It is a big Python pitfall.

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Not this time! 🙃

[awaelchli]

@hadim just letting you know :))

[pep8speaks]

Hello @awaelchli! Thanks for updating this PR.

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2020-04-20 05:24:13 UTC

[codecov]

Codecov Report

Merging #1534 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff           @@
##           master   #1534   +/-   ##
======================================
  Coverage      90%     90%           
======================================
  Files          68      68           
  Lines        3804    3804           
======================================
  Hits         3441    3441           
  Misses        363     363           

[Borda]

LGTM 🐰

[Borda]

LGTM 🐰

[hadim]

What??? I didn't know that! Thanks, @awaelchli for the catch.

[awaelchli]

Yep but no harm was done, because most users have anyway just one Trainer in their script and never got into trouble so far :) haha this is really a weird Python thing.
https://docs.python-guide.org/writing/gotchas/#mutable-default-arguments