(original) (raw)

On Thu, 14 Apr 2016 at 03:26 Victor Stinner <victor.stinner@gmail.com> wrote:

Le 14 avr. 2016 11:16 AM, "Serhiy Storchaka" <storchaka@gmail.com> a écrit :
\> A desirable but nonexistent feature is to write emails to authors of commits that broke buildbots. How hard to implement this?

Yeah I also had this idea since many years but buildbots were quite unstable. Maybe we should be more strict to consider a buildbot as stable?

Depending on how fancy we get with our infrastructure after we move to GitHub, we could theoretically end up with a PR-merging bot that can detect which commit broke things and report on the PR that did it (we well as report anywhere else we wanted to).

I propose to experiment sending notifications of failure to the authors of changes \*and\* to a new mailing list. I would subscribe to such list. An even safer starting point would be to only start with the mailing list.

FYI I'm connected to the #python-dev IRC channel which already contain these notifications. But I agree that mails are better.

Yeah, I'm one of those that doesn't sit on #python-dev due to the lack of a persistently connected machine, so an email would work better (unless we want to be trendy and write a bot for Slack/Skype/FB Messenger :).

> What are you think about backporting recent regrtest to 2.7? Most needed features to me are the -m and -G options.

Regrtest changed a lot in python 3.6 (new test.libregrtest library).
I suggest to start from python 3.5.

For -m: if it doesn't need to modify the unittest module, I agree.

I don't know -G option.

> Would be nice to add a feature for running every test in separate subprocess. This will isolate the effect of failed tests.

See my email :-) I proposed to modify -j1 to run tests in subrpocesses. I even mentionned my issue.

I suggest to use -jN on all buildbot, at least -j1.

Maybe -j2 is even better since many tests are waiting on IO or simple sleep.

Both ideas seems reasonable.