On Wed, Jun 6, 2018, 15:48 Victor Stinner, <vstinner@redhat.com> wrote:
2018-06-06 19:37 GMT+02:00 Brett Cannon <brett@python.org>:
My point is this is all software which we know isn't perfect and as long as there is a way to kick off a wedged/failed build again I don't think it's a hard blocker.
Ok, let me elaborate.
I'm trying to stabilize Travis CI, AppVeyor and buildbots for 1 month. I got some help on some issues, but sadly too few core devs are working on making CIs more stable, whereas CIs are blocking our workflow when they fail. It's annoying when a PR is blocked by a failure which is unrelated to the proposed change.
In 1 month, I already saw significant progress, but there are still a few open issues: https://mail.python.org/pipermail/python-dev/2018-June/153796.html
My point is that making VSTS mandatory adds more issues. I would prefer to have more time to fix existing *known* issues first. I'm not talking about VSTS issues in particular, but all CI known issues.
Fair enough!
In the meanwhile, AppVeyor seems stable again. Would it be possible to re-enable it in all stable branches?
It was never turned off, but if you mean make it required again then yes.