[MPlayer-cvslog] r36045 - in trunk: command.c configure libaf/af.c libmpcodecs/ve_lavc.c libmpcodecs/ve_x264.c libmpdemux/mf.c

[Diego Biurrun]

On Sat, Mar 23, 2013 at 02:01:14PM +0100, Reimar Döffinger wrote:
> On Sat, Mar 23, 2013 at 01:58:28PM +0100, Reimar Döffinger wrote:
> > On Sat, Mar 23, 2013 at 02:46:57PM +0200, Ivan Kalvachev wrote:
> > > On 3/23/13, diego <subversion at mplayerhq.hu> wrote:
> > > > Author: diego
> > > > Date: Sat Mar 23 01:50:51 2013
> > > > New Revision: 36045
> > > >
> > > > Log:
> > > > build: Add _BSD_SOURCE definition for strsep().
> > > >
> > > > Modified:
> > > >    trunk/command.c
> > > >    trunk/configure
> > > >    trunk/libaf/af.c
> > > >    trunk/libmpcodecs/ve_lavc.c
> > > >    trunk/libmpcodecs/ve_x264.c
> > > >    trunk/libmpdemux/mf.c
> > > 
> > > Why are you putting _BSD_SOURCE in each file and not in CPPFLAGS, like
> > > you did with the other visibility flags in r36009?
> > 
> > I we'd rather not by accident introduce new functions that require
> > _BSD_SOURCE
> > A different question would be why not add it to osdep/strsep.h,
> > however the answer there is that it won't work unless we include
> > that file always before the system headers, which is kind of messy.
> 
> However to enforce consistency maybe at least a
> #ifndef _BSD_SOURCE
> #error ...
> 
> in osdep/strsep.h would be a good idea?
> Though maybe the best thing to do would be to just rename
> our strsep implementation to mpstrsep and always use that,
> forbidding use of the system strsep.
> It would at least result in consistent behaviour and reduce
> configure complexity.

Why do we need it in the first place?  It was added in 2002 to fix a
lack of strsep() in Solaris.  Solaris 9 came out later in that year.
So this seems to be a hack for a decade-old proprietary operating
system.  Not worth fussing over or keeping IMO.

Diego