[3.7] Remove unused imports in tests (GH-14518) #14522
Conversation
(cherry picked from commit 8f4ef3b)
vstinner
requested review from
1st1,
asvetlov,
gpshead,
rhettinger and
a team
as code owners
|
Why is this being backported into a maintenance branch? What bug is this fixing? What bugs are being introduced by it? |
Oh sorry, I only explained it in the main PR (for master): Otherwise, automated backports will more likely fail with a dummy conflict. I prefer when the bot is able to automatically backport changes to 3.7 and 3.8 ;-) I'm trying to backport all bugfixes to 2.7, 3.7 and 3.8: all maintained branches. Note: I chose to not backport this change to 2.7, there are likely too many conflicts, and 2.7 is getting less and less changes. |
|
I don't want to make a big deal out of this but, In general, reducing the potential for future backport conflicts is not an acceptable reason for changing a maintenance branch:
To do otherwise is to jeopardize the stability and compatibility guarantees that we make to our downstream users (i.e. that they can upgrade from one micro release to any higher micro release without anything breaking), solely for our own convenience. Sure, there is not a hard and fast rule: factors like where a particular release is in its life cycle should be taken into account. So backporting such a change from master to a branch that is still in a pre-release phase (like 3.8 is at the moment) is one thing. But 3.7 has been released for a year and is halfway through its maintenance phase. We should not be planning to backport lots of changes to it at this point. And changes to tests are still changes to the codebase of a release. Changes to the tests can hide errors and make it more difficult to see regressions. What are the chances that some new bug has been introduced in the changes to over 40 files - for no functional benefit? |
|
@ned-deily: Ok, I created PR #14555 to revert it. I will merge it as soon as the CI tests pass. |
|
I reverted this change ;-) |
(cherry picked from commit 8f4ef3b)