CodSpeed Performance Report

Merging #12657 will not alter performance

_{Comparing dylwil3:tuple-comprehensions (6862cd3) with main (d6c6db5)}

Summary

✅ 32 untouched benchmarks

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

ℹ️ ecosystem check detected linter changes. (+7 -0 violations, +0 -0 fixes in 4 projects; 50 projects unchanged)

ruff check --no-cache --exit-zero --ignore RUF9 --output-format concise --preview --select ALL

+ airflow/providers/google/cloud/transfers/sql_to_gcs.py:259:41: C409 Unnecessary list comprehension passed to `tuple()` (rewrite as a generator)
+ dev/breeze/src/airflow_breeze/utils/packages.py:374:12: C409 Unnecessary list comprehension passed to `tuple()` (rewrite as a generator)
+ tests/_internals/capture_warnings.py:187:55: C409 Unnecessary list comprehension passed to `tuple()` (rewrite as a generator)
+ tests/always/test_example_dags.py:132:25: C409 Unnecessary list comprehension passed to `tuple()` (rewrite as a generator)

ruff check --no-cache --exit-zero --ignore RUF9 --output-format concise --preview

+ latch/functions/operators.py:106:27: C409 Unnecessary list comprehension passed to `tuple()` (rewrite as a generator)

ruff check --no-cache --exit-zero --ignore RUF9 --output-format concise --preview

+ tests/utils/test_uri.py:194:20: C409 Unnecessary list comprehension passed to `tuple()` (rewrite as a generator)

ruff check --no-cache --exit-zero --ignore RUF9 --output-format concise --preview

+ mypy/checkpattern.py:390:17: C409 Unnecessary list comprehension passed to `tuple()` (rewrite as a generator)

I'd argue this is only safe to flag for list comprehensions as set comprehensions will dedup and potentially reorder elements as well (not that you should rely on that nondeterm ordering but still).

Good point! Another thing I noticed is that the suggestion text doesn't make sense in the context of comprehensions. I'll fix both next time I'm in front of a computer.

Thanks!

Ok, should be good to go!

I find this kind of thing unfortunate, because tuple(i for i in range(10000)) is like 50% slower than tuple([i for i in range(10000)]).

There's no way to configure ruff to disambiguate between the two.
Edit: I should probably open a new issue #12912 :-)