11:39 <cmouse> uh

11:39 <cmouse> is there some way to split lambdas?

11:42 <cmouse> answering myself, maybe

12:32 <wzzrd> hi folks, struggling a bit here: i got buildbot to build my projects nicely, but only on GitPoller; I have Gitea push pull requests to BB, they show up in the list of recent changes, but no builds happen for PRs

12:32 <wzzrd> I'm sure this is a 'duh' thing, but any pointers greatly appreciated

13:22 <cmouse> wzzrd: do you have more than 1 scheduler that matches the changes?

16:01 <aakashjain> tardyp p12tic: what should be next steps for https://github.com/buildbot/buildbot/issues/5906 ? should e2fb0fb be reverted?

16:06 <tardyp> e2fb0fb fixed another regression

16:09 <tardyp> so I guess we should fix the two usecase properly

16:10 <tardyp> which may just require putting this line back: https://github.com/buildbot/buildbot/commit/e2fb0fb3514b99c7e42137a6b97eb6b876a6a97a#diff-e77ab9c0beb5448f34c444ea388a12754b60ec7ba820cc6a0550989f37f3bde4L607

16:11 <tardyp> aakashjain: could you try it?

16:12 <aakashjain> tardyp: sure.

16:13 <aakashjain> tardyp: you mean having self.realUpdateSummary() before as well (at both, old and new location), right?

16:13 <tardyp> yes

16:13 <tardyp> so that the summary is updated before the step finish event is sent

16:14 <aakashjain> tardyp: sure, trying right away, give me few minutes.

16:14 <tardyp> but we still keep the oportunity to update in case of exception in the last stage

16:21 <aakashjain> tardyp: that didn't help, still noticing the issue after adding back self.realUpdateSummary()

16:21 <aakashjain> tardyp: if I also add back yield self.realUpdateSummary.stop() after self.realUpdateSummary, then it works. (but that would be almost reverting e2fb0fb)

16:24 <aakashjain> tardyp: not sure if having these two lines at both old and new location is a problem (would that still fix the issue which e2fb0fb fixed).

16:25 <aakashjain> tardyp: or can we move self.master.data.updates.finishStep() below self.realUpdateSummary.stop()

16:27 <aakashjain> tardyp: verified that moving self.master.data.updates.finishStep below self.realUpdateSummary.stop() also fixes the issue.

16:28 <tardyp> its because updateSummary is debounced

16:28 <tardyp> so it waits one second before actually doing the update

16:29 <aakashjain> ok

16:29 <aakashjain> tardyp: : https://pastebin.com/AWnHYLFi (this is the patch which works)

16:29 <infobob> https://bpaste.net/TSVQ (repasted for aakashjain)

16:30 <tardyp> I dont think so, because the purpose of the original patch was to catchup any issues with this step

16:30 <tardyp> I think we need a flush() method on debouncer

16:30 <tardyp> which will just cancel the time and call invoke()

16:31 <tardyp> https://github.com/buildbot/buildbot/blob/master/master/buildbot/util/debounce.py#L85

16:31 <tardyp> actually flus

16:32 <tardyp> actually flush() can just be stop(), then start()

16:35 <aakashjain> tardyp: ok

16:36 <aakashjain> tardyp: and the original patch (e2fb0fb) fixes the issue specifically for new-style steps? or was it for old-style steps as well?

16:39 <tardyp> I think this is unrelated to step's style

16:43 <aakashjain> tardyp: ok

16:43 <aakashjain> tardyp: I am not very clear on the debouncer thing. Maybe you can tell me the specific changes, and I verify that out on my buildbot instance.

16:48 <tardyp> https://github.com/buildbot/buildbot/pull/5952/files

16:48 <tardyp> but its on main branch

17:36 <aakashjain> tardyp: Thanks for creating the PR. I applied it on my buildbot instance, and it works fine (after minor fixes which I commented in PR).

18:15 <tardyp> aakashjain: so p12tic actually agrees on your idea to delay the finishStep data update

18:15 <tardyp> so I switched the implementation

18:43 <aakashjain> tardyp: sounds good!