528681 - [Checkins] Module reports duplicate check-ins

Status: RESOLVED WORKSFORME

(Webtools Graveyard :: Mozbot, defect)

Description

[Tanner M. Young [:tmyoung]]

I think this is another "fun" encoding problem.

Frequently mozbot reports checkins twice.  It always happens when Dão commits, and I think it is because of the spelling of his name, the "ã":

[01:20:40]	http://hg.mozilla.org/releases/mozilla-1.9.2/rev/71f4671663c9 - Dão Gottwald - Bug 525651 - On Windows, tabbrowser tabs often lack legibility when a lightweight theme is selecte. r=rflint
[01:21:24]	Check-in: http://hg.mozilla.org/releases/mozilla-1.9.2/rev/71f4671663c9 - D?o Gottwald - Bug 525651 - On Windows, tabbrowser tabs often lack legibility when a lightweight theme is selecte. r=rflint

There are other duplicates that occur that may be unrelated, but I have never noticed a duplicate listing of one of my commits, and I never use non-ascii-type chars.

Comment 1

[Wolf [:wolf]]

Technically, this is a hg.mozilla.org problem.. there's 2 webheads, and one of them does not serve the feed with the right encoding, and the checkins (and rdf) modules, bases its duplicate filtering on having the same title. (which for checkins, is a combination of the url to the changeset, the name and the push comment..

Comment 2

[Tanner M. Young [:tmyoung]]

I was afraid it would be something like that.  What would be the cleanest way to fix it?  I wonder if we could use only the changeset url, because there should never be two changesets with the same id.  This way even if the wrong encoding is returned--which I have not seen before--there would be no duplicate entry.  Would that work?

I can check with the hg people to make sure that it is impossible to have changesets of the same id, but I would assume that is the case.

Comment 3

[Wolf [:wolf]]

Does this still occur?