Re: [RFC] On sending forge pull request discussions to the mailing list
Christophe Lyon <[email protected]>
| Newsgroups | gmane.comp.ai.sourceware.forge,gmane.comp.gcc.devel |
|---|---|
| Message-ID | <[email protected]> |
On 4/9/26 17:27, Claudio Bantaloukas wrote:
> On 08/04/2026 14:52, Ben Boeckel wrote:
>> On Tue, Apr 07, 2026 at 14:04:18 +0100, Richard Earnshaw (foss) wrote:
>>> I suspect the best solution would be if we could re-implement the
>>> Linaro bot as a runner. But there are security implications that
>>> would need to be investigate for that, since the runner would then be
>>> managed by the forge (sent jobs), but would be in a separate
>>> administrative/security domain.
>>
>> Hmm…I don't think that should be necessary. This is the API I'm thinking
>> of:
>>
>> https://forge.sourceware.org/api/swagger#/repository/
>> repoCreateStatus
>>
>> /repos/{owner}/{repo}/statuses/{sha}
>>
>> It should render near any native CI results as well.
>>
>
> Hi Ben,
> thank you for pointing this out. I completely missed this feature when
> looking at the API. (/facepalm)
>
> Indeed this gives us options:
> - some checks could be entirely hosted by external entities, using
> hardware and scripts that they control and they can simply send a status
> on the commit stating their process works with the build on that commit.
> - Ideally the majority of these checks would run on containers whose
> definitions are made available under contrib/ci-containers and the
> process can be started with a single command under contrib/ci-scripts or
> just a yaml in .forgejo/workflows as that would allow all developers to
> replicate the run locally and do something about it.
> - some checks could post a status on each commit (like the check that
> changelogs are correct and the formatting adheres to the guidelines) and
> the majority would run against the tip of the pull request
>
> It would be great if the linaro CI could post statuses rather than
> messages for:
> linaroci / arm / build
> linaroci / arm / check
> linaroci / aarch64 / build
> linaroci / aarch64 / check
>
> and perhaps write a full message only when a failure is detected.
I'm convinced it can be updated to do that, I'll check with them.
>
>> Thanks,
>>
>> --Ben