On Mon, Jul 11, 2011 at 10:40:15AM +0100, Måns Rullgård wrote: > Diego Biurrun <[email protected]> writes: > > > On Mon, Jul 11, 2011 at 12:47:53AM +0100, Måns Rullgård wrote: > >> Diego Biurrun <[email protected]> writes: > >> > >> > On Sun, Jul 10, 2011 at 11:44:53PM +0100, Måns Rullgård wrote: > >> >> Diego Biurrun <[email protected]> writes: > >> >> > >> >> > None of the compiled binaries use any of libpostproc's symbols except > >> >> > for trivial ones that print version information. Thus the dependency > >> >> > yields no benefit and it is preferable to drop it. > >> >> > --- > >> >> > Makefile | 6 ++++-- > >> >> > cmdutils.c | 2 -- > >> >> > 2 files changed, 4 insertions(+), 4 deletions(-) > >> >> > >> >> [...] > >> >> > >> >> > +SUBDIRS-$(CONFIG_POSTPROC) += postproc > >> >> > +SUBDIRS = $(SUBDIRS-yes) $(FFLIBS) > >> >> > >> >> This part feels a little weird. > >> > > >> > Do you have another suggestion? We use the -yes trick everywhere... > >> > >> That wasn't what I meant. > > > > You are speaking in riddles. It's unavoidable to work with two lists, > > since the list of libraries to link against is not identical to the > > list of libraries to install. Since the list of install libraries is > > a superset of the link libraries, it makes sense to me to construct the > > former out of the latter. > > That's not what I meant either. I'm talking about how you construct > that list I quoted. I just seems weird. Compare to other similar > lists.
So let us compare to the FFLIBS list, which is very directly related: FFLIBS-$(CONFIG_AVDEVICE) += avdevice FFLIBS := $(FFLIBS-yes) $(FFLIBS) This looks identical to me (modulo the ':'). The only weird thing I notice is you getting a weird feeling :) Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
