bpo-44682: Handle invalid arg to pdb's "commands" directive

[akulakov]

https://bugs.python.org/issue44682

[akulakov]

@iritkatriel Thanks for catching this! expanded the tests, fixed news entry & replied to one comment.

[akulakov]

@iritkatriel committed your suggestions.

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @iritkatriel for commit 9e61254 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

[akulakov]

I looked a bit into the failure, was not able to find anything definitive, but there were a few recent merged PRs with similar or more extensive changes, could the error be caused by one of them?

"D:\buildarea\pull_request.bolen-windows10\build\PCbuild\sqlite3.vcxproj" (default target) (35) ->
d:\buildarea\pull_request.bolen-windows10\build\externals\sqlite-3.35.5.0\sqlite3.c : error C2471: cannot update program database 'D:\buildarea\pull_request.bolen-windows10\build\PCbuild\obj\311amd64_Debug\sqlite3\vc141.pdb' [D:\buildarea\pull_request.bolen-windows10\build\PCbuild\sqlite3.vcxproj]
d:\buildarea\pull_request.bolen-windows10\build\externals\sqlite-3.35.5.0\sqlite3.c : fatal error C1051: program database file, 'D:\buildarea\pull_request.bolen-windows10\build\PCbuild\obj\311amd64_Debug\sqlite3\vc141.pdb', has an obsolete format, delete it and recompile [D:\buildarea\pull_request.bolen-windows10\build\PCbuild\sqlite3.vcxproj]

[iritkatriel]

It doesn't look related, maybe rebasing will help.

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @iritkatriel for commit c080ecb 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

[iritkatriel]

@ambv Any idea why the "AMD64 Windows10 PR" test is failing? We tried to rebase, but that didn't seem to help. Thanks.

[ambv]

@iritkatriel Currently that particular buildbot worker (bolen-windows10) is failing all builds with:

fatal error C1060: compiler is out of heap space

I contacted the buildbot owner as this looks like a buildbot failure (Windows 10 Pro PR buildbot is fine, as is Azure Pipelines, and Github Actions).

Unless urgent, I'd recommend waiting for the buildbot to return to a green state. I will let you know on this PR when this happens.

[ambv]

Well, the good news is that the Windows 10 buildbot is back and running, the owner notified me it was in a bad state due to Windows updates clogging up most of the RAM.

The bad news is that the Windows 8.1 Refleaks buildbot is red on this PR only. This looks weird to me since the tests passed, just the exit code is non-zero. In any case, I scheduled it to re-run: https://buildbot.python.org/all/#/buildrequests/315077

Unfortunately, refleaks are pretty slow to re-run so we might not be able to see the new result today.

[akulakov]

@ambv thanks for following up on this, no rush!

[akulakov]

There is one failure that seems to be unrelated AFAICT:

test test_concurrent_futures failed -- Traceback (most recent call last):
File "D:\buildarea\pull_request.ware-win81-release.refleak\build\lib\test\test_concurrent_futures.py", line 905, in test_idle_thread_reuse
self.assertEqual(len(executor._threads), 1)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: 2 != 1

[iritkatriel]

I agree, that doesn't seem related. @ambv any objections to merging this as it is?

[bedevere-bot]

@ambv: Please replace # with GH- in the commit message next time. Thanks!

[ambv]

Thanks for bearing with me on this. I'll look into this importlib weirdness. It's worrying it is reproducible but in any case it is unrelated to this PR.

[iritkatriel]

Thank you @ambv!

[akulakov]

@ambv No worries, thank you! @iritkatriel much thanks for reviewing!