Skip to content

Conversation

@fjetter
Copy link
Member

@fjetter fjetter commented Feb 27, 2025

xref #11736

This caches the tokens on the expression and restores calculated tokens after a pickle roundtrip. This helps us keep tokens consistent between client and scheduler.
This is safe to do since Expr are by definition immutable (since otherwise our singleton approach would fall apart anyhow)

@github-actions
Copy link
Contributor

github-actions bot commented Feb 27, 2025

Unit Test Results

See test report for an extended history of previous test failures. This is useful for diagnosing flaky tests.

      9 files  ±  0        9 suites  ±0   3h 25m 58s ⏱️ - 1m 54s
 17 820 tests + 15   16 604 ✅ + 13   1 214 💤 ±0  2 ❌ +2 
159 454 runs  +135  147 297 ✅ +134  12 155 💤  - 1  2 ❌ +2 

For more details on these failures, see this check.

Results for commit f4cd5ac. ± Comparison against base commit 5f61e42.

♻️ This comment has been updated with latest results.

dask/_expr.py Outdated
@functools.cached_property
def _name(self) -> str:
return self._funcname + "-" + _tokenize_deterministic(*self.operands)
if not self._token:
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The name is overwritten in a few places, is this an issue?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe. I'll check

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yeah, that's an issue

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

there are just 38 places this is overridden one way or another... no big deal!

@fjetter
Copy link
Member Author

fjetter commented Feb 27, 2025

I think I now broke out all tokenization from the actual naming. There are a few subclasses that overwrite the tokenization although I believe that should not be necessary.

@fjetter
Copy link
Member Author

fjetter commented Feb 27, 2025

One of the test failures is fixed by #11795 and the other one is a random weird thing of aioboto

@phofl phofl merged commit fe46cee into dask:main Feb 27, 2025
22 of 24 checks passed
@phofl
Copy link
Collaborator

phofl commented Feb 27, 2025

thanks

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants