svn commit: r381955 - head/graphics/gdal

[Alexey Dokuchaev]

On Mon, Mar 23, 2015 at 02:57:11PM +0100, Baptiste Daroussin wrote:
> On Sun, Mar 22, 2015 at 10:14:13PM -0500, Bryan Drewery wrote:
> > On 3/22/2015 8:58 PM, Alexey Dokuchaev wrote:
> > > On Sun, Mar 22, 2015 at 08:11:17PM +0000, Sunpoet Po-Chuan Hsieh wrote:
> > >> New Revision: 381955
> > >> URL: https://svnweb.freebsd.org/changeset/ports/381955
> > >> QAT: https://qat.redports.org/buildarchive/r381955/
> > >>
> > >> Log:
> > >>   - Use -lpthread instead of -pthread
> > > 
> > > Is -pthread going away anytime soon?  I'm a bit worried of more alike
> > > changing one to another; since they're identical right now on FreeBSD,
> > > I'd rather reduce amount of patching, leaving whatever is picked by
> > > upstream, unless -lpthread really should be preferred over -pthread
> > > for some reason I'm unaware of (and ready to learn about).
> > 
> > It's a pointless change. Tijl explains here
> > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=198116#c4
> 
> This is not pointless at all this is a long standing thing that should be
> cleaned out for a while, portmgr already gave long ago some blanket to some
> developers to remove that along with all usage of PTHREAD_LIBS. The reason
> is that it is confusing for maintainers, as if they take example on some of
> our existing ports they might be driven in the wrong direction.

Right; actually it was me who misread original commit diff, which actually
does the right thing by removing REINPLACE_CMD, and while LDFLAGS change
might seen gratuitous, both John and Baptiste are correct that if preferred
way now is -lpthread (which luckily everyone happens to agree at), all *our*
references (e.g. ports' Makefiles) to it should be updated.

My own concern is that if *upstream* sources (Makefiles) keep using -pthread
it's IMHO fine and should not be patched, but that's 1) arguable, and 2) has
nothing to do with r381955.  Sorry for the noise.

./danfe