Page MenuHomePhabricator

Gerrit bot posts duplicate changeset announcements
Closed, ResolvedPublic

Event Timeline

Nemo_bis raised the priority of this task from to Needs Triage.
Nemo_bis updated the task description. (Show Details)
Nemo_bis added a project: Phabricator.
Nemo_bis changed Security from none to None.
Nemo_bis subscribed.

As far as I can see, they are not duplicate: The first one (of the last three...) was posted on Tue, Dec 2, 12:42 AM, the second on Fri, Dec 5, 2:55 AM and the last one on Tue, Dec 9, 4:27 AM.

As far as I can see, they are not duplicate: The first one (of the last three...) was posted on Tue, Dec 2, 12:42 AM, the second on Fri, Dec 5, 2:55 AM and the last one on Tue, Dec 9, 4:27 AM.

They are duplicate in content, not in timestamp.

Aklapper triaged this task as Lowest priority.Dec 9 2014, 12:53 PM

I'm pretty sure this was the same behavior in Bugzilla so I don't consider this a bug and nothing to fix.

I'm pretty sure this was the same behavior in Bugzilla so I don't consider this a bug and nothing to fix.

Definitely not, gerritbot used to announce only newly-added changesets.

There are four comments in T42192 that stem from change 149293:

The first comment is from the change's first Patch Set. This Patch Set
has Bugzilla references in commit message. This is from before the
Bugzilla -> Phabricator switch. This comment looks sane.

The second comment is from switching from Bugzilla references to
Phabricator references in the commit message. The different gerrit
plugins are not coordinated, so the Gerrit->Phabricator plugin does
not know about the Gerrit->Bugzilla plugin and what it already reported.
Hence, the plugins connot know that “Bug: 40192” and “Bug: T42192
refer to the same thing. This is an edge case from the migration, and
I'd argue that fixing this is not worth the required time.

Note that he third and fourth comments are different from the second.
While the second command is about a patch set having been
uploaded, the third and fourth comments are about a Patch Set
(that was previously in draft state) has been published.

This comment-on-all-published-drafts behaviour has been around
since ever (also on Bugzilla), but I guess it never happened that
someone published more draft sets for a change with bug references
(or people did not care before :-) ).
Anyways it looks wrong.
Let's try to see if it's easy to fix :-)

Change 182751 had a related patch set uploaded (by QChris):
Make Gerrit only comment for published drafts that add new task references

https://gerrit.wikimedia.org/r/182751

Patch-For-Review

Change 182751 merged by Rush:
Make Gerrit only comment for published drafts that add new task references

https://gerrit.wikimedia.org/r/182751

QChris claimed this task.