svn commit: r428763 - in head/audio/fluidsynth: . files

Thomas Zander riggs at freebsd.org
Sat Dec 17 17:17:03 UTC 2016


On 17 December 2016 at 18:00, Mathieu Arnold <mat at freebsd.org> wrote:
> Le 17/12/2016 à 17:38, Thomas Zander a écrit :
>> On 17 December 2016 at 14:50, Mathieu Arnold <mat at freebsd.org> wrote:
>>> Le 17/12/2016 à 14:41, Thomas Zander a écrit :
>>>> Modified: head/audio/fluidsynth/Makefile
>>>> ==============================================================================
>>>> --- head/audio/fluidsynth/Makefile    Sat Dec 17 13:26:05 2016        (r428762)
>>>> +++ head/audio/fluidsynth/Makefile    Sat Dec 17 13:41:32 2016        (r428763)
>>>> @@ -13,7 +13,7 @@ COMMENT=    Real-time software synthesizer
>>>>  LICENSE=     LGPL20
>>>>  LICENSE_FILE=        ${WRKSRC}/COPYING
>>>>
>>>> -OPTIONS_DEFINE=      JACK ALSA DBUS LADSPA LASH PORTAUDIO PULSEAUDIO SNDFILE
>>>> +OPTIONS_DEFINE=      JACK ALSA DBUS LADSPA LASH PORTAUDIO PULSEAUDIO SNDFILE SNDIO
>>>>  OPTIONS_DEFAULT=     JACK
>>> This is missing a PORTREVISION bump.
>> Since when? The new SNDIO knob is disabled by default and the default
>> package remains unchanged.
>
> First, because, well, it is the rule[1]. You changed the options list,
> so the package can change, so it must be bumped.

Thanks for pointing to the doc. However, I specifically remember
discussions about NOT bumping in case the default package remains
unchanged, the goal being to avoid futile rebuild of packages. That
was the rationale for not bumping. But if it's the rule, sure, don't
want to break any laws here :-)

> Second, because if I had built this port before you added the option,
> and I had sndio installed, it would be linked with it but there would
> not be a dependency recorded. But if I build it after this option is
> added, it is not liked with SNDIO any more. Which makes a different
> package, which is a big reason for bumping PORTREVISION.

That's incorrect. The patch brings in not only the cmake args but the
whole sndio output module. fluidsynth before this commit would never
link to sndio. So this would have not been a problem at all.

Best regards
Riggs


More information about the svn-ports-head mailing list