[pre-commit.ci] pre-commit autoupdate

[pre-commit-ci]

updates:

• github.com/psf/black: 24.2.0 → 24.4.2
• github.com/astral-sh/ruff-pre-commit: v0.2.2 → v0.4.3
• github.com/asottile/pyupgrade: v3.15.1 → v3.15.2

[lgray]

@martindurant what has broken here?

[martindurant]

I swear I replied to this already...

There seem to be two separate things going on:

• one in dask-expr/blockwise, original exception
  AttributeError: 'HighLevelGraph' object has no attribute '_all_external_keys'
• the "spans" exception which is hist-specific, possibly since v1.4.1

[lgray]

Probably this PR got closed and re-opened and I still haven't had time to go chase down what happened in the HLG interface to remove _all_external_keys.

[lgray]

So one weird thing here is that the AttributeException that's causing that failure should be caught in get_all_external_keys which is what is being called. This must be some pytest configuration thing?

[martindurant]

I believe it's spread across dask and dask-expr, and probably to do with the latter being extended to (some) array situations. This will be super annoying!

[martindurant]

• the version of hist doesn't seem to matter
• Somehow, during compute, the dask-dataframe is still a dask-dataframe while evaluating tasks; it should be the pandas-dataframe associated with one partition.

[martindurant]

bisect says it's dask/dask#10977 (dask-expr got turned on, so this is super unhelpful)

[lgray]

Can we turn off dask-expr in our tests? Our primary use case is dask-awkward and boost-histograms.

[martindurant]

Like this, I think. I assume this could be in the dask config files

DASK_DATAFRAME__QUERY_PLANNING=False

[martindurant]

dask.config.set("dataframe.query-planning", False)

[martindurant]

OK, well it's green, but eventually we want dask-histogram to work happily with dask-expr too.