svn commit: r330906 - head/multimedia/vcdpad

[Alexey Dokuchaev]

On Sat, Oct 19, 2013 at 06:39:46PM +0200, Baptiste Daroussin wrote:
> On Sat, Oct 19, 2013 at 04:36:48PM +0000, Alexey Dokuchaev wrote:
> > On Sat, Oct 19, 2013 at 04:30:33PM +0000, Baptiste Daroussin wrote:
> > > New Revision: 330906
> > > URL: http://svnweb.freebsd.org/changeset/ports/330906
> > > 
> > > [...]
> > > +post-install:
> > > +	${STRIP_CMD} ${STAGEDIR}${PREFIX}/bin/${PORTNAME}
> > 
> > How do you know if installed binaries should be post-stripped?  Is there
> > automated check involved?  What if, for any reason, binary indeed must be
> > installed unstripped?
> 
> make stage warns about unstripped binaries if you have DEVELOPERS=YES in your
> make.conf,

I suppose it should read DEVELOPER, not DEVELOPER_S_, right?

> if you need the binaries unstripped then just do not strip them, those works
> stripped properly, btw STRIP_CMD respects WITH_DEBUG.

<pessimistic mode on>
I understand; I was worrying about rare cases when stripping the binary (not
for debugging) actually renders it unusable in certain circumstances.  Don't
get me wrong, I also prefer to have installed binaries normally stripped, but
I vaguely recall there were cases, albeit rarely seen, when upstream actually
intended unstripped installs, for a reason.  I just hope that these sorts of
automatic suggestions would not result in massive commits that would result
in broken installations.  From what I see how easily folks follow portlint's
sometimes misguided advice, it could become problem.
</pessimistic mode off>

./danfe