Page MenuHomePhabricator

diff_checker should ignore new files when porting from compat
Closed, DeclinedPublic

Event Timeline

Why shouldn't we fix them instead? Should we ignore flake8 errors, too?

flake8 errors could be disabled.

The reason is that porting a whole file from compat all lines are tested but normally I fix differences to the core framework only. Using the new styles any code diff tools would show me a lot of changed lines and it is more difficult to verify the important changes. Maybe this is a minor prio request or we shouldn't care about this because most of scripts and modules are ported already.

Can we ignore the jenkins-bot's verification by manually changing the status to +1 or +2 and force the merge, or will it prevent the merge? (I've never tried)

If not, or if that's not a good enough solution, should we:

  • Ignore any newly added file
  • in the scripts directory
  • when the word "compat" is in the commit message

? Is this a good enough heuristic?

@Dalba: I guess we could ignore bot's verification but have to submit the change manually to merge it to the repository but we shouldn't do it except in some unusal cases (maybe the bots don't work properly)

your suggested solution would be enough I think. or look for a pattern "# Automatically ported from compat branch by script"

Xqt triaged this task as Low priority.Jan 7 2018, 11:54 AM

Which scripts are left to port from compat? Has anyone asked for them? If not, we're importing more maintenance work without any clear benefit.

According to my comment in T186465.

offtopic, but I agree with @jayvdb: We should port only those someone requested or [are/want to be used] on some project. I ported ndashredir (and rewritten it a little bit), because I need it for some tasks on Czech Wikipedia. And I would suggest the same to others.