Use __class__ attribute for 5% speed gain.

[StephenBrown2]

Change Summary

Updates https://github.com/samuelcolvin/pydantic/tree/use-class-attribute on top of current master. It isn't 2x faster, which may be why the branch hasn't been merged yet, but it is faster, about 5% on average, as demonstrated below (avg=117.592μs/iter on master vs avg=111.204μs/iter on use-class-attribute):

# On branch master
❯ BENCHMARK_REPEATS=100 make benchmark-pydantic
python benchmarks/run.py pydantic-only
                                pydantic time=0.674s, success=50.10%
                                ...
                                pydantic time=0.670s, success=50.10%
                                pydantic best=0.647s, avg=0.706s, stdev=0.111s

                                pydantic best=107.908μs/iter avg=117.592μs/iter stdev=18.569μs/iter version=1.4a1

❯ git checkout use-class-attribute
Switched to branch 'use-class-attribute'
Your branch is up to date with 'mine/use-class-attribute'.

❯ BENCHMARK_REPEATS=100 make benchmark-pydantic
python benchmarks/run.py pydantic-only
                                pydantic time=0.824s, success=50.10%
                                ...
                                pydantic time=0.660s, success=50.10%
                                pydantic best=0.637s, avg=0.667s, stdev=0.032s

                                pydantic best=106.111μs/iter avg=111.204μs/iter stdev=5.374μs/iter version=1.4a1

Related issue number

None

Checklist

• Unit tests for the changes exist
• Tests pass on CI and coverage remains at 100%
• Documentation reflects the changes where applicable
• changes/<pull request or issue id>-<github username>.md file added describing change
  (see changes/README.md for details)

[samuelcolvin]

interesting, I couldn't get a confirmed improvement.

Did you compile pydantic for the benchmarks?

[StephenBrown2]

Yes, I got the same relative results after compiling (and recompiling):

❯ git checkout master && make clean build-cython benchmark-pydantic && git checkout use-class-attribute && make clean build-cython benchmark-pydantic
master:                 pydantic best=75.329μs/iter avg=85.574μs/iter stdev=11.919μs/iter version=1.4a1
use-class-attribute:    pydantic best=74.781μs/iter avg=81.755μs/iter stdev=8.646μs/iter version=1.4a1

[samuelcolvin]

@StephenBrown2 could you fix tests?

[codecov]

Codecov Report

Merging #1294 into master will not change coverage by %.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #1294   +/-   ##
=======================================
  Coverage   99.89%   99.89%           
=======================================
  Files          21       21           
  Lines        3713     3713           
  Branches      731      731           
=======================================
  Hits         3709     3709           
  Misses          2        2           
  Partials        2        2           

Impacted Files	Coverage Δ
pydantic/dataclasses.py	100.00% <100.00%> (ø)
pydantic/env_settings.py	100.00% <100.00%> (ø)
pydantic/error_wrappers.py	100.00% <100.00%> (ø)
pydantic/fields.py	100.00% <100.00%> (ø)
pydantic/json.py	100.00% <100.00%> (ø)
pydantic/mypy.py	100.00% <100.00%> (ø)
pydantic/networks.py	100.00% <100.00%> (ø)
pydantic/schema.py	99.44% <100.00%> (ø)
pydantic/typing.py	100.00% <100.00%> (ø)
pydantic/utils.py	98.92% <100.00%> (ø)
... and 1 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fcf5478...5572acb. Read the comment docs.

[StephenBrown2]

Tests fixed!

[samuelcolvin]

I think codecov is broken, I'll try and fix that and fix coverage later.

[StephenBrown2]

Coverage actually updated, no change. The only line that's not covered is still if allow_names and field_name.isidentifier(): in utils.py