Regression tests with -w should only re-run affected test methods, not the entire file

[ambv]

BPO	44708
Nosy	@ambv, @pablogsal, @miss-islington
PRs	bpo-44708: Only re-run test methods that match names of previously failing test methods #27287 [3.10] bpo-44708: Only re-run test methods that match names of previously failing test methods (GH-27287) #27290 [3.9] bpo-44708: Only re-run test methods that match names of previously failing test methods (GH-27287) #27293

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/ambv'
closed_at = <Date 2021-07-22.21:18:18.842>
created_at = <Date 2021-07-22.14:32:29.824>
labels = ['3.11', 'tests', '3.9', '3.10', 'performance']
title = 'Regression tests with -w should only re-run affected test methods, not the entire file'
updated_at = <Date 2021-07-22.21:18:18.842>
user = 'https://github.com/ambv'

bugs.python.org fields:

activity = <Date 2021-07-22.21:18:18.842>
actor = 'lukasz.langa'
assignee = 'lukasz.langa'
closed = True
closed_date = <Date 2021-07-22.21:18:18.842>
closer = 'lukasz.langa'
components = ['Tests']
creation = <Date 2021-07-22.14:32:29.824>
creator = 'lukasz.langa'
dependencies = []
files = []
hgrepos = []
issue_num = 44708
keywords = ['patch']
message_count = 4.0
messages = ['397985', '397994', '398001', '398007']
nosy_count = 3.0
nosy_names = ['lukasz.langa', 'pablogsal', 'miss-islington']
pr_nums = ['27287', '27290', '27293']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'resource usage'
url = 'https://bugs.python.org/issue44708'
versions = ['Python 3.9', 'Python 3.10', 'Python 3.11']

[ambv]

When the CPython test suite re-runs a flaky failed test, it doesn’t actually re-run just that failed test but the entire test file. The most common case for flaky tests is when networking, threading, or multiprocessing is involved. Frustratingly those are some of the largest test files we have. Re-running them takes a lot of time.

Instead of re-running the entire file, regrtest should only re-run what actually failed.

NOTE: I added 3.10 and 3.9 to this issue even though it's not a pure bugfix. The reason is that this can visibly speed up CI for open pull requests.

(Of course, it would be best to avoid flaky tests altogether and we’re working on that, but re-running what failed is a pragmatic stopgap that is necessary in times of distributed CI that we don’t fully control.)

[ambv]

New changeset f1afef5 by Łukasz Langa in branch 'main':
bpo-44708: Only re-run test methods that match names of previously failing test methods (GH-27287)
f1afef5

[ambv]

New changeset 9608719 by Miss Islington (bot) in branch '3.10':
bpo-44708: Only re-run test methods that match names of previously failing test methods (GH-27287) (GH-27290)
9608719

[ambv]

New changeset 0f55d21 by Łukasz Langa in branch '3.9':
[3.9] bpo-44708: Only re-run test methods that match names of previously failing test methods (GH-27287) (GH-27293)
0f55d21