Declare istr with __slots__ = () in _multidict_py.py

[jonathandung]

What do these changes do?

Optimizes memory usage by preventing a dictionary to store attributes to be created on every istr instance.

Are there changes in behavior for the user?

Users can no longer set attributes on the keys of the Python version of case-insensitive multidicts, mutable or otherwise.

Related issue number

Fixes #1360.

Checklist

• I think the code is well written
• Unit tests for the changes exist
• Documentation reflects the changes

[Vizonex]

If your going to use slots I would recommend trying to add __istr_identitiy__ to it. While I like your idea for speeding up pure python the tests reflect the issue.

AttributeError: 'istr' object attribute '__istr_identity__' is read-only

A Way you could slove this one is to try

__slots__ = ("__istr_identity__",)

[jonathandung]

Wait, but it's impossible to define a default value for a slot by setting it on the class, since that overwrites the slot. I changed the relevant logic a bit.

[codspeed-hq]

Merging this PR will degrade performance by 21.09%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 11 regressed benchmarks
✅ 231 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_iterate_multidict_keys[ci-py]	245.1 µs	346.6 µs	-29.29%
❌	test_iterate_multidict[ci-py]	246.2 µs	346.1 µs	-28.86%
❌	test_iterate_multidict_items[ci-py]	264.4 µs	364.3 µs	-27.42%
❌	test_keys_view_less_or_equal[ci-py]	285.2 µs	383.3 µs	-25.59%
❌	test_keys_view_less[ci-py]	290.8 µs	389.8 µs	-25.4%
❌	test_items_view_less[ci-py]	335 µs	437.1 µs	-23.35%
❌	test_items_view_less_or_equal[ci-py]	332.6 µs	433.6 µs	-23.28%
❌	test_multidict_popitem_str[ci-py]	675.8 µs	781.6 µs	-13.54%
❌	test_keys_view_sub[ci-py]	1 ms	1.2 ms	-11.92%
❌	test_keys_view_or[ci-py]	1.1 ms	1.3 ms	-9.83%
❌	test_items_view_or[ci-py]	1.2 ms	1.4 ms	-9.75%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing jonathandung:master (2de45be) with master (c63dfaf)¹}

Footnotes

1. No successful run was found on master (a4271e8) during the generation of this report, so c63dfaf was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[Vizonex]

@codspeedbot fix this regression

[jonathandung]

I'm not familiar with this benchmarking framework, and I don't understand what it means by "different runtime environments detected".

[jonathandung]

I have done all I can, and I think the regression was caused by CodSpeed running on a different virtual CPU. I'm not sure though.

[Vizonex]

@webknjaz Any thoughts on this PR?