On Tue, Jul 15, 2025 at 3:32 AM Patrick Steinhardt <ps@xxxxxx> wrote:
>
> On Tue, Jul 15, 2025 at 01:56:44AM -0700, Carlo Arenas wrote:
> > On Tue, Jul 15, 2025 at 1:46 AM Patrick Steinhardt <ps@xxxxxx> wrote:
> > > On Mon, Jul 14, 2025 at 09:55:27PM -0400, Eli Schwartz wrote:
> > > > On 7/13/25 1:48 PM, Carlo Marcelo Arenas B
> > > > > diff --git a/meson_options.txt b/meson_options.txt
> > > > > index e7f768df24..f63ff32556 100644
> > > > > --- a/meson_options.txt
> > > > > +++ b/meson_options.txt
> > > > > @@ -45,7 +45,7 @@ option('gitweb', type: 'feature', value: 'auto',
> > > > > description: 'Build Git web interface. Requires Perl.')
> > > > > option('iconv', type: 'feature', value: 'auto',
> > > > > description: 'Support reencoding strings with different encodings.')
> > > > > -option('pcre2', type: 'feature', value: 'enabled',
> > > > > +option('pcre2', type: 'feature', value: 'auto',
> > > >
> > > > This part is fine. We shouldn't default-fail if it isn't found, when we
> > > > can't expect it to be universally available.
> > >
> > > Agreed. I guess tha only reason why I picked "enabled" here is because
> > > we also got a wrapper in "subprojects/". But with this new workaround in
> > > place I agree that it is sensible to switch to "auto".
> >
> > AFAIK the "wrapper" fallback still kicks in when the feature is "auto"
>
> It does, yes. But with 'auto' as default it means that we're free to
> disable PCRE2 if we have detected a broken PCRE2 dependency.
My bad, I was mistaken and indeed auto doesn't fallback to the wrap, so
this will likely regress in windows if it is not invoked with `-Dpcre2=enabled`
Carlo
