llvm dev - Oct 2019 - Phabricator picking up downstream commits from Github forks of llvm-project?

That appears to have been pushed to
refs/am/changes/bbc4c751f01bb6f649942d3cf3b504a87c9019c8_swift/master-next

Does someone (perhaps someone from apple) know what that is? Was it pushed
there, rather than the swift fork, by mistake?

On Tue, Oct 29, 2019 at 6:12 PM Anton Korobeynikov via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> Someone pushed these changes to LLVM repo. See e.g.
> https://github.com/llvm/llvm-project/commit/8910c5c78688
>
> On Tue, Oct 29, 2019 at 12:33 PM Alex L via llvm-dev
> <llvm-dev at lists.llvm.org> wrote:
> >
> > Hi,
> >
> > It looks like LLVM's Phabricator started picking up downstream
commits
> in Apple's fork of llvm-project (github.com/apple/llvm-project), and is
> creating notification events about all the old downstream commits, e.g.
> >
> > https://reviews.llvm.org/rG8910c5c786886f17a75bd142fa967932ca3f54c1
> > https://reviews.llvm.org/rGb03469c2d72621e1cccfeeaef719692600c075f4
> >
> > This seems like a bug. Can this be disabled?
> >
> > Thanks,
> > Alex
> > _______________________________________________
> > LLVM Developers mailing list
> > llvm-dev at lists.llvm.org
> > https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
>
>
> --
> With best regards, Anton Korobeynikov
> Department of Statistical Modelling, Saint Petersburg State University
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20191029/2188513d/attachment.html>

Oh, this explains it! Unfortunately one of our engineers made a mistake,
and pushed the ref to wrong remote while resolving a merge conflict on
https://github.com/apple/llvm-project (pushed to
https://github.com/llvm/llvm-project instead of
https://github.com/apple/llvm-project). I just deleted the ref from the
https://github.com/llvm/llvm-project. It's weird that Phabricator decided
to pick this up, since it's not in the usual refs/heads namespace.
Hopefully now it will stop going through those commits.

We will work on improving the process on our end to avoid mistakes like
this in the future.

Thanks,
Alex

On Tue, 29 Oct 2019 at 15:50, James Y Knight <jyknight at google.com>
wrote:
> That appears to have been pushed to
> refs/am/changes/bbc4c751f01bb6f649942d3cf3b504a87c9019c8_swift/master-next
>
> Does someone (perhaps someone from apple) know what that is? Was it pushed
> there, rather than the swift fork, by mistake?
>
> On Tue, Oct 29, 2019 at 6:12 PM Anton Korobeynikov via llvm-dev <
> llvm-dev at lists.llvm.org> wrote:
>
>> Someone pushed these changes to LLVM repo. See e.g.
>> https://github.com/llvm/llvm-project/commit/8910c5c78688
>>
>> On Tue, Oct 29, 2019 at 12:33 PM Alex L via llvm-dev
>> <llvm-dev at lists.llvm.org> wrote:
>> >
>> > Hi,
>> >
>> > It looks like LLVM's Phabricator started picking up downstream
commits
>> in Apple's fork of llvm-project (github.com/apple/llvm-project),
and is
>> creating notification events about all the old downstream commits, e.g.
>> >
>> >
https://reviews.llvm.org/rG8910c5c786886f17a75bd142fa967932ca3f54c1
>> >
https://reviews.llvm.org/rGb03469c2d72621e1cccfeeaef719692600c075f4
>> >
>> > This seems like a bug. Can this be disabled?
>> >
>> > Thanks,
>> > Alex
>> > _______________________________________________
>> > LLVM Developers mailing list
>> > llvm-dev at lists.llvm.org
>> > https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>
>>
>>
>> --
>> With best regards, Anton Korobeynikov
>> Department of Statistical Modelling, Saint Petersburg State University
>> _______________________________________________
>> LLVM Developers mailing list
>> llvm-dev at lists.llvm.org
>> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20191029/53f5df72/attachment.html>

October 30, 2019 5:58 AM, "Alex L via llvm-dev" <llvm-dev at
lists.llvm.org> wrote:
> Oh, this explains it! Unfortunately one of our engineers made a mistake,
and pushed the ref to
> wrong remote while resolving a merge conflict on
https://github.com/apple/llvm-project (pushed to
> https://github.com/llvm/llvm-project instead of
https://github.com/apple/llvm-project). I just
> deleted the ref from the https://github.com/llvm/llvm-project. It's
weird that Phabricator decided
> to pick this up, since it's not in the usual refs/heads namespace.
Hopefully now it will stop going
> through those commits.
> 
> We will work on improving the process on our end to avoid mistakes like
this in the future.
What's a good way to avoid these mistakes?

Sameer.

On Wed, 30 Oct 2019, Sameer Sahasrabuddhe via llvm-dev wrote:
> October 30, 2019 5:58 AM, "Alex L via llvm-dev" <llvm-dev at
lists.llvm.org> wrote:
>
>> Oh, this explains it! Unfortunately one of our engineers made a
mistake, and pushed the ref to
>> wrong remote while resolving a merge conflict on
https://github.com/apple/llvm-project (pushed to
>> https://github.com/llvm/llvm-project instead of
https://github.com/apple/llvm-project). I just
>> deleted the ref from the https://github.com/llvm/llvm-project. It's
weird that Phabricator decided
>> to pick this up, since it's not in the usual refs/heads namespace.
Hopefully now it will stop going
>> through those commits.
>> 
>> We will work on improving the process on our end to avoid mistakes like
this in the future.
>
> What's a good way to avoid these mistakes?
One way that I use to avoid accidentally pushing to the wrong repo, is to 
make sure that the "origin" remote (which is set as the upstream for
the
master branch, where pushes would go implicitly) is set up via an url that 
is (to me) read-only.

When dealing with github, I don't have a stored authenticated session for 
https, so to me, https://github.com/llvm/llvm-project is read-only (or, if 
I try to push there, it'd prompt me for username/password, which I'd 
notice). When pushing to github, I use ssh as transport, with urls like 
git at github.com:llvm/llvm-project, and add this as a separate remote that I 
only use for pushing. That forces me to explicitly spell out e.g. "git 
push <read-write-remote-name> master" to push there.

If access to github via https is authenticated already, one can choose to 
set the origin remote to a git://github.com/llvm/llvm-project url instead, 
which always is read only.

None of this helps against accidentally doing "git push <accidentally 
wrong remote name> <other branch>" though.

// Martin