On Tue, Mar 27, 2012 at 6:06 AM, Tim Starling <[email protected]> wrote:
> I could not find a recommendation to use --amend in the Gerrit manual.
> The manual says that it is possible to submit subsequent commits to
> the same change by just adding a Change-Id footer to the commit
> message. That seems to be the reason for the copy button next to the
> Change-Id on the change page.
>
> When I tried to do a test push containing several commits which
> depended on each other, but with the same Change-Id, Gerrit rejected
> it. But I'm not sure if that's a configuration issue or if it just
> expected them to be done in separate pushes.
>
> If it was possible to follow the Gerrit manual in this way, and submit
> non-amending followup commits with the same Change-Id, then we'd have
> the comment grouping advantages without the code review disadvantages.
>

Yes, that works. And actually, I think it's a little bit easier. It's a
little more "manual" that trying git one-liners, but it seems to at
least *always* work (and I've used it a couple of times to fix
changes with totally botched history).

>> An unrelated disadvantage of "stacked changes" in Gerrit (i.e. changes
>> that depend on other unmerged changes) is that if B.1 (change B
>> patchset 1) depends on A.1, and someone amends A.1 later to produce
>> A.2, B.1 will still depend on A.1 and will need to be rebased to
>> depend on A.2 instead. I've done this before with a stack of four
>> commits (amended B and had to rebase C and D), and I can tell you it's
>> not fun. I think I've figured out a trick for it now (use an
>> interactive rebase from the top of the stack), but it's not intuitive
>> and I've never tried it.
>
> Tell people to not amend their commits then.
>

Also making sure people don't have unrelated commits showing up as
dependencies makes the process much easier. If two things aren't
related--don't relate them!

-Chad

_______________________________________________
Wikitech-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Reply via email to