On Thu, 3 Sept 2026 at 13:15, Philippe Mathieu-Daudé
<[email protected]> wrote:
>
> On 3/9/26 12:04, Paolo Bonzini wrote:
> > The QEMU community has broad consent to not accept assisted-by and similar
> > trailers, and much less fake co-authored-bys, independent of any AI policy.
> > Implement this in checkpatch.pl already, so that there is another way for
> > contributors to discover the AI policy.

> > +# Reject trailers that credit an AI agent.
> > +             if ($realfile =~ /^$/ &&
> > +                 ($line =~ /🤖/ ||
>
> Few years ago Peter was concerned about using non-ASCII because
> it was not displaying correctly on his terminal:
> https://lore.kernel.org/qemu-devel/cafeaca-cjwce9k+tqq-vdrevzzpv_+cxar1j9qfr_ivjtj3...@mail.gmail.com/
> (that was using Ubuntu 22.04 which we don't support anymore).
> Should we still worry about non-ASCII in the repository? Having
> a quick look at the OS versions we support it seems all of them
> are ready to handle UTF-8 chars. Anyway I saw various utf-8
> emoji symbols in gitlab issues, and it is likely someone will
> paste one along with the report when filling commit description
> for a fix, so they'll end up displayed in a terminal via git-log.

An emoji in checkpatch because we're specifically looking for it
is probably be fine. Though I guess it's implicitly assuming the
user's locale is UTF-8, which I'm not sure it will be. We seem tu
use \xNN escape sequences elsewhere in this script where we
need to look for oddball non-ASCII, and that would be safer I guess.

In the thread above I was mostly pushing back because that
was a suggestion of using an emoji in *an executable filename*,
which is going to be exposed to many more people than a string
in a script to be run by developers.

thanks
-- PMM

Reply via email to