gh-118761: Improve import time by lazy import of warnings

[donbarbos]

• Improve import time modules by lazy import warnings.
• Remove unused warnings and struct from Lib/ctypes/_layout.py.

My hyperfine benchmarks:

First experiment in Main branch

➜ cpython git:(43e0240213) ✗ hyperfine --warmup 11"./python -c 'import codeop'""./python -c 'import zipimport'""./python -c 'import ctypes'""./python -c 'import hmac'""./python -c 'import importlib'""./python -c 'import logging'""./python -c 'import pkgutil'""./python -c 'import pydoc'""./python -c 'import rlcompleter'""./python -c 'import ssl'""./python -c 'import tempfile'""./python -c 'import traceback'""./python -c 'import urllib'""./python -c 'import xml'" Benchmark 1: ./python -c 'import codeop' Time (mean ± σ): 24.6ms± 1.0ms[User: 21.8 ms, System: 2.7 ms] Range (min … max): 23.4ms… 32.9ms112 runs 
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 2: ./python -c 'import zipimport'
 Time (mean ± σ): 22.7ms± 0.8ms[User: 20.0 ms, System: 2.6 ms]
 Range (min … max): 21.8ms… 28.6ms128 runs
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 3: ./python -c 'import ctypes'
 Time (mean ± σ): 27.4ms± 0.5ms[User: 24.1 ms, System: 3.2 ms]
 Range (min … max): 26.5ms… 29.3ms108 runs
Benchmark 4: ./python -c 'import hmac'
 Time (mean ± σ): 27.2ms± 0.5ms[User: 23.8 ms, System: 3.3 ms]
 Range (min … max): 26.3ms… 28.9ms108 runs
Benchmark 5: ./python -c 'import importlib'
 Time (mean ± σ): 23.1ms± 0.9ms[User: 20.3 ms, System: 2.6 ms]
 Range (min … max): 22.1ms… 31.3ms122 runs
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 6: ./python -c 'import logging'
 Time (mean ± σ): 51.9ms± 0.6ms[User: 47.8 ms, System: 4.0 ms]
 Range (min … max): 50.6ms… 53.4ms56 runs
Benchmark 7: ./python -c 'import pkgutil'
 Time (mean ± σ): 52.2ms± 0.9ms[User: 48.3 ms, System: 3.8 ms]
 Range (min … max): 50.4ms… 54.8ms58 runs
Benchmark 8: ./python -c 'import pydoc'
 Time (mean ± σ): 89.2ms± 1.7ms[User: 83.4 ms, System: 5.6 ms]
 Range (min … max): 87.2ms… 95.0ms34 runs
Benchmark 9: ./python -c 'import rlcompleter'
 Time (mean ± σ): 61.8ms± 1.1ms[User: 57.0 ms, System: 4.6 ms]
 Range (min … max): 60.1ms… 65.8ms49 runs
Benchmark 10: ./python -c 'import ssl'
 Time (mean ± σ): 47.6ms± 1.8ms[User: 43.1 ms, System: 4.3 ms]
 Range (min … max): 45.1ms… 57.6ms63 runs
Benchmark 11: ./python -c 'import tempfile'
 Time (mean ± σ): 47.2ms± 1.1ms[User: 42.9 ms, System: 4.3 ms]
 Range (min … max): 45.2ms… 49.9ms63 runs
Benchmark 12: ./python -c 'import traceback'
 Time (mean ± σ): 43.5ms± 1.2ms[User: 39.8 ms, System: 3.5 ms]
 Range (min … max): 41.7ms… 46.8ms65 runs
Benchmark 13: ./python -c 'import urllib'
 Time (mean ± σ): 25.9ms± 1.7ms[User: 22.8 ms, System: 2.9 ms]
 Range (min … max): 22.5ms… 34.3ms128 runs
Benchmark 14: ./python -c 'import xml'
 Time (mean ± σ): 26.2ms± 0.9ms[User: 23.5 ms, System: 2.6 ms]
 Range (min … max): 24.5ms… 29.2ms114 runs
Summary
 ./python -c 'import zipimport' ran
1.02 ± 0.05 times faster than ./python -c 'import importlib'
1.08 ± 0.06 times faster than ./python -c 'import codeop'
1.14 ± 0.09 times faster than ./python -c 'import urllib'
1.16 ± 0.06 times faster than ./python -c 'import xml'
1.20 ± 0.05 times faster than ./python -c 'import hmac'
1.21 ± 0.05 times faster than ./python -c 'import ctypes'
1.92 ± 0.09 times faster than ./python -c 'import traceback'
2.08 ± 0.09 times faster than ./python -c 'import tempfile'
2.10 ± 0.11 times faster than ./python -c 'import ssl'
2.29 ± 0.09 times faster than ./python -c 'import logging'
2.30 ± 0.09 times faster than ./python -c 'import pkgutil'
2.72 ± 0.11 times faster than ./python -c 'import rlcompleter'
3.93 ± 0.16 times faster than ./python -c 'import pydoc'

Second experiment in Main branch

➜ cpython git:(43e0240213) ✗ hyperfine --warmup 11"./python -c 'import codeop'""./python -c 'import zipimport'""./python -c 'import ctypes'""./python -c 'import hmac'""./python -c 'import importlib'""./python -c 'import logging'""./python -c 'import pkgutil'""./python -c 'import pydoc'""./python -c 'import rlcompleter'""./python -c 'import ssl'""./python -c 'import tempfile'""./python -c 'import traceback'""./python -c 'import urllib'""./python -c 'import xml'" Benchmark 1: ./python -c 'import codeop' Time (mean ± σ): 24.5ms± 1.6ms[User: 21.7 ms, System: 2.8 ms] Range (min … max): 23.5ms… 39.7ms119 runs 
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 2: ./python -c 'import zipimport'
 Time (mean ± σ): 22.8ms± 1.0ms[User: 20.4 ms, System: 2.3 ms]
 Range (min … max): 21.7ms… 30.3ms129 runs
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 3: ./python -c 'import ctypes'
 Time (mean ± σ): 27.6ms± 0.9ms[User: 24.5 ms, System: 2.9 ms]
 Range (min … max): 26.5ms… 34.3ms105 runs
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 4: ./python -c 'import hmac'
 Time (mean ± σ): 27.1ms± 0.6ms[User: 23.7 ms, System: 3.3 ms]
 Range (min … max): 26.2ms… 28.8ms108 runs
Benchmark 5: ./python -c 'import importlib'
 Time (mean ± σ): 23.9ms± 1.2ms[User: 21.4 ms, System: 2.3 ms]
 Range (min … max): 22.0ms… 26.4ms126 runs
Benchmark 6: ./python -c 'import logging'
 Time (mean ± σ): 59.0ms± 2.9ms[User: 54.6 ms, System: 4.3 ms]
 Range (min … max): 56.5ms… 73.0ms52 runs
Benchmark 7: ./python -c 'import pkgutil'
 Time (mean ± σ): 61.2ms± 3.9ms[User: 56.1 ms, System: 5.0 ms]
 Range (min … max): 55.4ms… 68.4ms53 runs
Benchmark 8: ./python -c 'import pydoc'
 Time (mean ± σ): 114.4ms± 3.4ms[User: 106.5 ms, System: 7.7 ms]
 Range (min … max): 111.3ms… 122.7ms26 runs
Benchmark 9: ./python -c 'import rlcompleter'
 Time (mean ± σ): 78.8ms± 1.2ms[User: 72.5 ms, System: 6.1 ms]
 Range (min … max): 76.7ms… 81.0ms38 runs
Benchmark 10: ./python -c 'import ssl'
 Time (mean ± σ): 58.8ms± 0.9ms[User: 53.4 ms, System: 5.3 ms]
 Range (min … max): 57.4ms… 61.6ms50 runs
Benchmark 11: ./python -c 'import tempfile'
 Time (mean ± σ): 59.8ms± 1.8ms[User: 54.2 ms, System: 5.5 ms]
 Range (min … max): 57.4ms… 69.4ms50 runs
Benchmark 12: ./python -c 'import traceback'
 Time (mean ± σ): 54.8ms± 1.0ms[User: 50.3 ms, System: 4.4 ms]
 Range (min … max): 52.4ms… 56.9ms56 runs
Benchmark 13: ./python -c 'import urllib'
 Time (mean ± σ): 29.5ms± 0.6ms[User: 26.2 ms, System: 3.1 ms]
 Range (min … max): 28.3ms… 31.9ms100 runs
Benchmark 14: ./python -c 'import xml'
 Time (mean ± σ): 29.7ms± 1.1ms[User: 26.7 ms, System: 3.0 ms]
 Range (min … max): 28.2ms… 38.3ms102 runs
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Summary
 ./python -c 'import zipimport' ran
1.05 ± 0.07 times faster than ./python -c 'import importlib'
1.08 ± 0.08 times faster than ./python -c 'import codeop'
1.19 ± 0.06 times faster than ./python -c 'import hmac'
1.21 ± 0.07 times faster than ./python -c 'import ctypes'
1.29 ± 0.06 times faster than ./python -c 'import urllib'
1.30 ± 0.07 times faster than ./python -c 'import xml'
2.40 ± 0.11 times faster than ./python -c 'import traceback'
2.58 ± 0.12 times faster than ./python -c 'import ssl'
2.58 ± 0.17 times faster than ./python -c 'import logging'
2.62 ± 0.14 times faster than ./python -c 'import tempfile'
2.68 ± 0.20 times faster than ./python -c 'import pkgutil'
3.45 ± 0.16 times faster than ./python -c 'import rlcompleter'
5.01 ± 0.26 times faster than ./python -c 'import pydoc'

First experiment in PR branch

➜ cpython git:(issue-118761) ✗ hyperfine --warmup 11 "./python -c 'import codeop'" "./python -c 'import zipimport'" "./python -c 'import ctypes'" "./python -c 'import hmac'" "./python -c 'import importlib'" "./python -c 'import logging'" "./python -c 'import pkgutil'" "./python -c 'import pydoc'" "./python -c 'import rlcompleter'" "./python -c 'import ssl'" "./python -c 'import tempfile'" "./python -c 'import traceback'" "./python -c 'import urllib'" "./python -c 'import xml'" Benchmark 1: ./python -c 'import codeop' Time (mean ± σ): 23.5 ms ± 1.0 ms [User: 21.0 ms, System: 2.5 ms] Range (min … max): 22.6 ms … 33.2 ms 121 runs 
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might helptouse the '--warmup'or'--prepare' options.
Benchmark2: ./python -c 'import zipimport'
Time (mean ± σ): 22.6 ms ± 0.8 ms [User: 20.3 ms, System: 2.2 ms]
Range (min … max): 21.6 ms … 29.5 ms 127 runs
Warning: Statistical outliers were detected. Consider re-running this benchmarkon a quiet systemwithoutany interferences from other programs. It might helptouse the '--warmup'or'--prepare' options.
Benchmark3: ./python -c 'import ctypes'
Time (mean ± σ): 27.3 ms ± 0.5 ms [User: 24.3 ms, System: 3.0 ms]
Range (min … max): 26.4 ms … 29.0 ms 107 runs
Benchmark4: ./python -c 'import hmac'
Time (mean ± σ): 26.2 ms ± 0.6 ms [User: 23.3 ms, System: 2.9 ms]
Range (min … max): 25.3 ms … 27.7 ms 108 runs
Benchmark5: ./python -c 'import importlib'
Time (mean ± σ): 22.9 ms ± 1.3 ms [User: 20.6 ms, System: 2.3 ms]
Range (min … max): 22.0 ms … 35.4 ms 125 runs
Warning: Statistical outliers were detected. Consider re-running this benchmarkon a quiet systemwithoutany interferences from other programs. It might helptouse the '--warmup'or'--prepare' options.
Benchmark6: ./python -c 'import logging'
Time (mean ± σ): 51.3 ms ± 0.8 ms [User: 47.3 ms, System: 3.9 ms]
Range (min … max): 49.8 ms … 53.7 ms 58 runs
Benchmark7: ./python -c 'import pkgutil'
Time (mean ± σ): 51.1 ms ± 0.8 ms [User: 47.3 ms, System: 3.8 ms]
Range (min … max): 49.5 ms … 53.6 ms 58 runs
Benchmark8: ./python -c 'import pydoc'
Time (mean ± σ): 85.8 ms ± 2.5 ms [User: 80.5 ms, System: 5.2 ms]
Range (min … max): 84.4 ms … 99.8 ms 35 runs
Warning: Statistical outliers were detected. Consider re-running this benchmarkon a quiet systemwithoutany interferences from other programs. It might helptouse the '--warmup'or'--prepare' options.
Benchmark9: ./python -c 'import rlcompleter'
Time (mean ± σ): 60.6 ms ± 0.8 ms [User: 55.8 ms, System: 4.6 ms]
Range (min … max): 59.2 ms … 63.9 ms 49 runs
Benchmark10: ./python -c 'import ssl'
Time (mean ± σ): 45.4 ms ± 1.7 ms [User: 41.3 ms, System: 4.0 ms]
Range (min … max): 43.9 ms … 58.1 ms 65 runs
Warning: Statistical outliers were detected. Consider re-running this benchmarkon a quiet systemwithoutany interferences from other programs. It might helptouse the '--warmup'or'--prepare' options.
Benchmark11: ./python -c 'import tempfile'
Time (mean ± σ): 46.1 ms ± 1.7 ms [User: 41.8 ms, System: 4.2 ms]
Range (min … max): 44.3 ms … 58.2 ms 65 runs
Warning: Statistical outliers were detected. Consider re-running this benchmarkon a quiet systemwithoutany interferences from other programs. It might helptouse the '--warmup'or'--prepare' options.
Benchmark12: ./python -c 'import traceback'
Time (mean ± σ): 42.4 ms ± 0.8 ms [User: 38.7 ms, System: 3.4 ms]
Range (min … max): 40.4 ms … 44.3 ms 70 runs
Benchmark13: ./python -c 'import urllib'
Time (mean ± σ): 22.9 ms ± 0.6 ms [User: 20.4 ms, System: 2.4 ms]
Range (min … max): 22.0 ms … 24.8 ms 122 runs
Benchmark14: ./python -c 'import xml'
Time (mean ± σ): 24.6 ms ± 1.4 ms [User: 22.1 ms, System: 2.4 ms]
Range (min … max): 22.2 ms … 33.0 ms 126 runs
Summary
 ./python -c 'import zipimport' ran
1.01 ± 0.04 times faster than ./python -c 'import urllib'
1.01 ± 0.07 times faster than ./python -c 'import importlib'
1.04 ± 0.06 times faster than ./python -c 'import codeop'
1.09 ± 0.07 times faster than ./python -c 'import xml'
1.16 ± 0.05 times faster than ./python -c 'import hmac'
1.21 ± 0.05 times faster than ./python -c 'import ctypes'
1.88 ± 0.07 times faster than ./python -c 'import traceback'
2.01 ± 0.10 times faster than ./python -c 'import ssl'
2.04 ± 0.10 times faster than ./python -c 'import tempfile'
2.26 ± 0.08 times faster than ./python -c 'import pkgutil'
2.27 ± 0.09 times faster than ./python -c 'import logging'
2.68 ± 0.10 times faster than ./python -c 'import rlcompleter'
3.80 ± 0.17 times faster than ./python -c 'import pydoc'

Second experiment in PR branch

➜ cpython git:(issue-118761) ✗ hyperfine --warmup 11"./python -c 'import codeop'""./python -c 'import zipimport'""./python -c 'import ctypes'""./python -c 'import hmac'""./python -c 'import importlib'""./python -c 'import logging'""./python -c 'import pkgutil'""./python -c 'import pydoc'""./python -c 'import rlcompleter'""./python -c 'import ssl'""./python -c 'import tempfile'""./python -c 'import traceback'""./python -c 'import urllib'""./python -c 'import xml'" Benchmark 1: ./python -c 'import codeop' Time (mean ± σ): 23.7ms± 1.0ms[User: 21.0 ms, System: 2.6 ms] Range (min … max): 22.7ms… 31.0ms122 runs 
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 2: ./python -c 'import zipimport'
 Time (mean ± σ): 22.6ms± 0.5ms[User: 20.1 ms, System: 2.4 ms]
 Range (min … max): 21.7ms… 24.0ms126 runs
Benchmark 3: ./python -c 'import ctypes'
 Time (mean ± σ): 27.6ms± 0.5ms[User: 24.7 ms, System: 2.8 ms]
 Range (min … max): 26.5ms… 28.9ms104 runs
Benchmark 4: ./python -c 'import hmac'
 Time (mean ± σ): 26.5ms± 1.0ms[User: 23.2 ms, System: 3.2 ms]
 Range (min … max): 25.4ms… 33.1ms111 runs
Benchmark 5: ./python -c 'import importlib'
 Time (mean ± σ): 23.0ms± 0.7ms[User: 20.6 ms, System: 2.4 ms]
 Range (min … max): 22.1ms… 25.5ms121 runs
Benchmark 6: ./python -c 'import logging'
 Time (mean ± σ): 51.5ms± 0.9ms[User: 47.2 ms, System: 4.2 ms]
 Range (min … max): 49.9ms… 54.6ms58 runs
Benchmark 7: ./python -c 'import pkgutil'
 Time (mean ± σ): 51.9ms± 1.4ms[User: 47.8 ms, System: 4.0 ms]
 Range (min … max): 49.7ms… 58.4ms57 runs
Benchmark 8: ./python -c 'import pydoc'
 Time (mean ± σ): 88.1ms± 2.2ms[User: 82.1 ms, System: 5.8 ms]
 Range (min … max): 85.1ms… 93.3ms34 runs
Benchmark 9: ./python -c 'import rlcompleter'
 Time (mean ± σ): 63.6ms± 3.5ms[User: 58.6 ms, System: 4.9 ms]
 Range (min … max): 59.4ms… 76.9ms38 runs
Benchmark 10: ./python -c 'import ssl'
 Time (mean ± σ): 50.6ms± 1.5ms[User: 46.2 ms, System: 4.3 ms]
 Range (min … max): 48.9ms… 60.0ms60 runs
Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
Benchmark 11: ./python -c 'import tempfile'
 Time (mean ± σ): 50.7ms± 0.8ms[User: 45.7 ms, System: 4.8 ms]
 Range (min … max): 49.1ms… 53.1ms59 runs
Benchmark 12: ./python -c 'import traceback'
 Time (mean ± σ): 54.3ms± 3.5ms[User: 49.3 ms, System: 4.9 ms]
 Range (min … max): 46.8ms… 71.1ms63 runs
Benchmark 13: ./python -c 'import urllib'
 Time (mean ± σ): 30.0ms± 0.6ms[User: 26.9 ms, System: 3.0 ms]
 Range (min … max): 28.4ms… 32.0ms97 runs
Benchmark 14: ./python -c 'import xml'
 Time (mean ± σ): 29.9ms± 0.7ms[User: 26.7 ms, System: 3.1 ms]
 Range (min … max): 28.4ms… 32.1ms100 runs
Summary
 ./python -c 'import zipimport' ran
1.02 ± 0.04 times faster than ./python -c 'import importlib'
1.05 ± 0.05 times faster than ./python -c 'import codeop'
1.17 ± 0.05 times faster than ./python -c 'import hmac'
1.22 ± 0.04 times faster than ./python -c 'import ctypes'
1.32 ± 0.04 times faster than ./python -c 'import xml'
1.33 ± 0.04 times faster than ./python -c 'import urllib'
2.24 ± 0.08 times faster than ./python -c 'import ssl'
2.24 ± 0.06 times faster than ./python -c 'import tempfile'
2.28 ± 0.06 times faster than ./python -c 'import logging'
2.30 ± 0.08 times faster than ./python -c 'import pkgutil'
2.40 ± 0.16 times faster than ./python -c 'import traceback'
2.82 ± 0.17 times faster than ./python -c 'import rlcompleter'
3.90 ± 0.13 times faster than ./python -c 'import pydoc'

• Issue: Improve import time of various stdlib modules #118761

[picnixz]

Thanks for doing this. I did it for hmac locally since I'm implementing the HACL* hmac nut it's good to see it done here.

[picnixz]

Please revert the whitespaces and unrelated new lines changes (unless the codeowners tell you that you can keep the changes though)

[bedevere-app]

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

[AA-Turner]

We should verify that each change has non-noise benefits rather than changing everything at once in an omnibus PR. @donbarbos also please present your motivating statistics in a more readable manner, I expect to see a comparison of each module against baseline, rather than four large log outputs.

A

[jaraco]

I ran a test for importlib.resources, and this change if anything makes things worse:

 cpython issue-118761 🐚 hyperfine --warmup 11 "./python.exe -c 'import importlib.resources'" Benchmark 1: ./python.exe -c 'import importlib.resources' Time (mean ± σ): 26.6 ms ± 0.5 ms [User: 20.9 ms, System: 3.8 ms] Range (min … max): 25.4 ms … 28.7 ms 101 runs cpython issue-118761 [1] 🐚 git checkout @$(git merge-base HEAD main) HEAD is now at 0fef47e5bbd gh-55454: Add IMAP4 IDLE support to imaplib (#122542) cpython 0fef47e5bbd 🐚 hyperfine --warmup 11 "./python.exe -c 'import importlib.resources'" Benchmark 1: ./python.exe -c 'import importlib.resources' Time (mean ± σ): 25.9 ms ± 0.6 ms [User: 21.4 ms, System: 3.8 ms] Range (min … max): 24.8 ms … 27.6 ms 98 runs 

Sounds like we should revert those changes from this PR.

[AA-Turner]

I'll close this PR as the contributor has opened some smaller individual changes.

A