svn commit: r318646 - head/graphics/scale2x

Alexey Dokuchaev danfe at FreeBSD.org
Wed Jul 10 06:51:47 UTC 2013


On Wed, May 22, 2013 at 01:27:37AM +0000, Alexey Dokuchaev wrote:
> On Tue, May 21, 2013 at 03:30:05PM +0200, Baptiste Daroussin wrote:
> > On Tue, May 21, 2013 at 06:59:44AM +0000, Alexey Dokuchaev wrote:
> > > On Tue, May 21, 2013 at 06:17:27AM +0000, Martin Wilke wrote:
> > > > New Revision: 318646
> > > > URL: http://svnweb.freebsd.org/changeset/ports/318646
> > > > 
> > > > +OPTIONS_DEFINE=	DOCS EXAMPLES
> > > > +OPTIONS_DEFINE_i386=	MMX
> > > >  
> > > > [...]
> > > >  
> > > > +.if ${PORT_OPTIONS:MMMX}
> > > > +CPPFLAGS+=	-DHAVE_MMX
> > > > +.endif
> > > 
> > > Do you think it would make sense to enable MMX option by default
> > > depending on ${MACHINE_CPU:Mmmx}?
> > 
> > That will not fit the builders it works properly when building locally
> > but not in "package factories" except if I miss something.
> 
> Hmm, perhaps you're right.  I'll see if something can be tuned in bpm to
> ensure that package factories will produce packages with conservative
> defaults, yet allowing ports users to automatically pick suitable
> optimizations.

On the second thought, actually it looks like everything is fine.  With
empty /make.conf (or __MAKE_CONF=/dev/null), which is what I presume is
true for package factories, MACHINE_CPU is just "i486", which is exactly
the safe defaults.  Users can set CPUTYPE?= in their make.conf to get
augmented list of features.

That said, I keep thinking that MMX should be enabled by default if
MACHINE_CPU lists it; package builders will produce i486'ish packages,
no?  (Ditto for SSE, SSE2 in others ports.)

./danfe


More information about the svn-ports-all mailing list