CQ/PreCQ rejections should notify both the Owner and 'Commit-Ready'-er |
||||||||||
Issue description
(Not exactly sure where to file this. It's also possible this has been discussed and I didn't find the "WAI" denotation.)
Steps:
1. Developer A uploads change
2. Developer B really wants to see the change land
3. Developer B reviews and CQ+1's the change
4. Developer A {dies,goes-to-sleep,leaves the team,goes on vacation,has a spiritual epiphany and ditches work}
5. Change gets rejected by PreCQ / CQ
6. Email notification for step 5 gets sent to Developer A
What's expected:
7. Email notification for step 5 gets sent to Developer B
8. Developer B happily notices, and gets a chance to either fix errors to help push the change along, or else just CQ+1 again
9. Return to step 5 ;)
What happens:
7. No other email notification is sent
8. Developer A doesn't read his/her email
9. Developer B doesn't notice the change got rejected
10. Awesome bugfixes and cool features get delayed
Is my expectation reasonable?
,
Sep 26 2016
I don't know if everybody would be equally interested in this. I wonder if Gerrit has a per-user option to make it more noisy by email for users that want that.
,
Sep 26 2016
I'm curious, why don't you think people wouldn't be interested in this? I thought CQ-ready basically means "I want to see this land." And I figured a feedback loop of "sorry, I can't do that for you Dave" is a natural part of that process. Regarding noise: well, Gerrit's already noisy enough, but I've written email filters for the kinds of messages I don't want to see. Seeing too much seems better than seeing too little... But yeah, a per-user option would satisfy my request, if it existed.
,
Sep 27 2016
@4: I agree with that. NOTE also that long ago starring an issue would give you such notifications. It doesn't seem to any more.
,
Oct 12 2016
Marking as "won't fix" as we don't have bandwidth for this. Feel free to check with Gerrit team (jrn@ or dborowitz@) for input on this (we don't think they use CR bug)
,
Oct 12 2016
@7: Hi Autumn, Thanks for the response. Does the same recommendation apply for getting the 'star' feature to do what it used to, as Doug suggested? That would be a good generic backup solution actually, and it sounds more like a regression than a feature request.
,
Oct 13 2016
Is there a 'star' feature in gerrit? If so, yeah, would be nice solution. Our team doesn't own Gerrit though, so any regressions should be brought up with the gerrit-on-borg team.
,
Oct 28 2016
Team bandwidth may not be an issue, if this looks OK: https://chromium-review.googlesource.com/404684 Given the above (we're specifically marking these messages with "Owner"), then I'm not sure this is something the GoB team would fix...unless you expect them to differentiate "Reviewers" and "Starred" in APIs like this. I'm not actually sure what the "Starred" feature is supposed to do at all. If that doesn't look good, then I guess I'll see what I can get at http://go/gob-bug ...
,
Oct 28 2016
Filed this: b/32505336 I'll see if there's a more palatable alternative to CL:404684.
,
Nov 28 2016
Got some feedback on b/32505336 FYI. --- BTW, I noticed this while looking a little deeper: https://bugs.chromium.org/p/chromium/issues/detail?id=281547#c8 "It seems difficult to implement with GoB. What's easy to do is to send email only to CL owner. Would it be enough? I think in most cases an issue owner is the one who sets CQ+1..." So, the $subject bug was a conscious regression.
,
Dec 16 2016
Looks like we should be able to use the recently-implemented NotifyInfo entity for the ReviewInput action: https://gerrit-review.googlesource.com/Documentation/rest-api-changes.html#review-input https://gerrit-review.googlesource.com/Documentation/rest-api-changes.html#notify-info So I think the syntax is something like notify_details=TO:account_id ? I'm not very good with JSON. Anyway, I'm told "it will still take a while until this is rolled out to production." So I guess we wait.
,
Mar 20 2017
I believe GoB has supported the new API for a while now. Here's my attempt to use it. I could only test by hacking similar queries locally (I can't test the 'on_behalf_of' approach, as my account doesn't have permissions): https://chromium-review.googlesource.com/#/c/457169/
,
Mar 30 2018
,
Mar 30 2018
,
Jun 29 2018
Seems reasonable and should be a very change. Right now everyone gets notified of CL's landing but the CQ+1 person does not get notified of failures.
,
Jul 2
Well, I had this ready: https://chromium-review.googlesource.com/c/chromiumos/chromite/+/457169 I don't really have any desire to codify/unify the existing design, so I guess it's stalled. |
||||||||||
►
Sign in to add a comment |
||||||||||
Comment 1 Deleted