Please see:
https://sourceforge.net/p/cjam/activity/?page=0&limit=100
The commits look like they were committed by someone else besides who actually committed them. I remember running into a similar issue before and the issue was that the commit didn't identify the author email address correctly and ended up showing the the closest account that matched as the committer. I can't remember process to prove that is the case. Please have someone look into this issue.
Thanks
SourceForge Support
Suggested from IRC Regular (wjp)
Commit.author_url calls User.by_email_address calls EmailAddress.canonical, which will return nobody@example.com for invalid addresses, which gets passed to query.find(), possibly returning a random user with email address set to nobody@example.com?
More info from IRC support channel
aditsu
6:14 JBarrett11: there's also something weird at https://sourceforge.net/p/cjam/code/ci/2a0112fd06d417a56f514306feab960df5e85c2a/ - that commit only modified 1 file, nothing added
JBarrett11
6:14 hmm, odd
aditsu
6:15 .hgtags is duplicated apparently
JBarrett11
6:15 you made that commit just a couple of days ago?
aditsu
6:15 yeah
6:16 it's also one of the ones listed with the wrong user
6:16 it looks perfectly fine in my local repository
Looks like the mercurial commits in question just have a username, no email address associated with them. And our
EmailAddress.canonical
defaults to 'nobody@example.com' which happened to match with an existing user.We should change how we handle commits without email addresses. And maybe also not default to 'nobody@example.com'. (would have to assess impact of an exception or other return value)
Hi, I'm the affected user. There's nothing special about those commits, as I never put my email address in there, and I never had this kind of problem before.
The only thing that's different is that the servers were having some serious problems while I was pushing those commits.
I can try to push some more stuff and see if it happens again. If it does, it means you broke something on your side in the last month or two.
Ok I pushed a new commit and this time it credited "dSoft®". There's definitely something wrong going on now and it wasn't happening a couple of months ago.
[forge:site-support:#9707]
https://sourceforge.net/p/forge/site-support/9707/
and
[forge:site-support:#9644]
https://sourceforge.net/p/forge/site-support/9644/
Closed #724.
ib/7823
Changed
EmailAddress.canonical
to return None and fixed all the places which it affects.forge-classic also uses it in two places, but always with valid emails (sf aliases), so should not be any problems there.
Also created [#7833] for bug I discovered while working on this.
Related
Tickets:
#7833Last edit: Igor Bondarenko 2015-02-12
Glenn Randers-Pehrson <glennrp at users.sourceforge.net>
is another format to test and make sure is handled well. (From https://sourceforge.net/p/forge/site-support/9644/)