Maxim Cournoyer <[email protected]> writes:

> Hi,
>
> Tomas Volf <[email protected]> writes:
>
>> Carlo Zancanaro <[email protected]> writes:
>>
>>> On Thu, May 14 2026, Tomas Volf wrote:
>>>> I have encountered bug with the hook we install for adding Change-Id
>>>> trailer.  It for some reason inserts it as a first line of the commit
>>>> message instead of at the end.
>
> Would you mind to share how to reproduce the issue? It isn't obvious
> to me.

Right, sorry, that was in the initial bug report.  Copy&pasting here:

--8<---------------cut here---------------start------------->8---
$ git commit -m '--- test' --allow-empty
[master f07a209fb1] Change-Id: I02675f1053c80563402f1acddd0a64a537f4da55 --- 
test
$ git log -1 --format=%B
Change-Id: I02675f1053c80563402f1acddd0a64a537f4da55
--- test
--8<---------------cut here---------------end--------------->8---

Notice that Change-Id line is *before* `--- test' line.

>
> [...]
>
>>> I'm not sure why we're limiting ourselves to git versions and features
>>> from 2017.
>>
>> Given that the commit-msg hook has this header
>>
>> # From Gerrit Code Review 3.11.1.
>> #
>> # Part of Gerrit Code Review (https://www.gerritcodereview.com/)
>> #
>> # Copyright (C) 2009 The Android Open Source Project
>>
>> I would not rule out the option that Maxim just added it without
>> considering this particular detail.  I took the liberty of CC-ing him,
>> maybe he could answer this bit.
>
> It's, as the copyright suggests, a script borrowed from the Gerrit
> project :-).  It hasn't been modified.  About the ways we call it, if it
> resolves a clear issue, we do not need to constrain ourselves to an old
> git version the same the gerrit project might; using the git version
> which is in our current master tree would be fine, I think.

Great, since adding the --no-divider seems to do the trick, that could
be a fix.  At the same time, since we are just setting a single trailer
to a random value, it might be worth it to simplify the script a lot.

Tomas

-- 
There are only two hard things in Computer Science:
cache invalidation, naming things and off-by-one errors.



Reply via email to