Re: Precommits broken?

Brilliant. That seems like a very clean solution that we can implement today. I'll get started; thanks for the idea Andrew! On Thu, Jun 14, 2018 at 2:15 PM Udi Meiri wrote: > +1 for separate jobs if it gets us faster to pre-commit filtering > > On Thu, Jun 14, 2018 at 11:22 AM Kenneth Knowles

Re: Precommits broken?

+1 for separate jobs if it gets us faster to pre-commit filtering On Thu, Jun 14, 2018 at 11:22 AM Kenneth Knowles wrote: > I like Andrew's solution. Just totally separate jobs for automatic and > manual. > > Kenn > > On Thu, Jun 14, 2018 at 9:56 AM Lukasz Cwik wrote: > >> That seems like a

Re: Precommits broken?

I like Andrew's solution. Just totally separate jobs for automatic and manual. Kenn On Thu, Jun 14, 2018 at 9:56 AM Lukasz Cwik wrote: > That seems like a pretty good interim solution. > > On Thu, Jun 14, 2018 at 9:53 AM Andrew Pilloud > wrote: > >> If you always run one job for automated and

Re: Precommits broken?

If you always run one job for automated and another job for manual you wouldn't need to remember two trigger phrases. The automated jobs don't even need trigger phrases. As long as the status contexts are the same github users never have to know they are two separate jobs. Andrew On Thu, Jun 14,

Re: Precommits broken?

I thought of that as well but would find it annoying that I would need to remember two sets of triggers, the ones for the automated jobs and the ones for the manual runs. If we re-use the same precommit trigger phrase, we would get two runs (automated and manual) of effectively the same thing for

Re: Precommits broken?

Might there be a third option of creating a different jenkins job for PR change and manual triggers? It would clutter up the jenkins interface a bit, but they could both post status to the same commitStatusContext on Github, so no one would notice there. Andrew On Wed, Jun 13, 2018 at 11:14 PM

Re: Precommits broken?

Having submitted a patch to the ghprb-plugin repo before, I think that regretfully option (b) is probably the right decision here given that it's unlikely to get accepted, merged, released, and to have Infra update the plugin in under a week. On Wed, Jun 13, 2018 at 10:42 PM Scott Wegner wrote:

Re: Precommits broken?

Indeed, I was going to send out an email about pre-commit filtering, but we've already found some kinks and may need to revert it. The change was submitted in PR#5611 [1] and enables Jenkins triggering to only run pre-commits based on modified files. However, Udi noticed that this also prevents

Re: Precommits broken?

Precommit filter is a really coool optimization! -Rui On Wed, Jun 13, 2018 at 5:21 PM Andrew Pilloud wrote: > Ah, so this is intended and I didn't break anything? Cool! Sorry for the > false alarm, looks like a great build optimization! > > Andrew > > On Wed, Jun 13, 2018 at 5:06 PM

Re: Precommits broken?

Ah, so this is intended and I didn't break anything? Cool! Sorry for the false alarm, looks like a great build optimization! Andrew On Wed, Jun 13, 2018 at 5:06 PM Yifan Zou wrote: > Probably due to the precommit filter applied in #5611 > ? > > On Wed,

Re: Precommits broken?

Probably due to the precommit filter applied in #5611 ? On Wed, Jun 13, 2018 at 5:02 PM Andrew Pilloud wrote: > Looks like statuses got posted between me writing this email and sending > it. Still wondering why the python and go jobs appear to be

Re: Precommits broken?

Looks like statuses got posted between me writing this email and sending it. Still wondering why the python and go jobs appear to be missing? Andrew On Wed, Jun 13, 2018 at 5:00 PM Andrew Pilloud wrote: > Recent PRs don't appear to be running all the precommits, and success > status isn't

Precommits broken?

Recent PRs don't appear to be running all the precommits, and success status isn't being pushed to PRs. Anyone know what is going on? See: https://github.com/apache/beam/pull/5592 https://github.com/apache/beam/pull/5622 Andrew