On Sat, Oct 20, 2012 at 07:47:51PM +0100, Måns Rullgård wrote: > Diego Biurrun <di...@biurrun.de> writes: > > On Sat, Oct 20, 2012 at 06:46:58PM +0100, Måns Rullgård wrote: > >> Diego Biurrun <di...@biurrun.de> writes: > >> > >> > $(FATE_WMA_ENCODE): CMP = stddev > >> > $(FATE_WMA_ENCODE): REF = > >> > $(SAMPLES)/audio-reference/luckynight_2ch_44kHz_s16.wav > >> > > >> > -FATE_SAMPLES_AVCONV += $(FATE_WMA_ENCODE) > >> > +FATE_SAMPLES_AVCONV-$(call ENCDEC, WMAV1, ASF) += fate-wmav1-encode > >> > +FATE_SAMPLES_AVCONV-$(call ENCDEC, WMAV2, ASF) += fate-wmav2-encode > >> > fate-wma-encode: $(FATE_WMA_ENCODE) > >> > >> This makes the fate-wma-encode target fail is one of the encoders is > >> disabled. > > > > Yes, this is an issue; not one restricted to this patch or patchset though. > > The VC-1 tests, for instance, suffer from a similar issue. > > No, they don't. The fate-vc1 target will fail if the VC1 decoder is > disabled, but in that case running it is completely pointless anyway > since _all_ the tests depend on that.
OK. What I mean is that there is a simple way to make fate-vc1 not error out: FATE_VC1-$(CONFIG_VC1_DEMUXER) += fate-vc1_sa00040 FATE_SAMPLES_AVCONV-$(CONFIG_VC1_DECODER) += $(FATE_VC1-yes) ---> FATE_VC1-$(call DEMDEC, VC1, VC1) += fate-vc1_sa00040 FATE_SAMPLES_AVCONV += $(FATE_VC1-yes) fate-vc1: $(FATE_VC1-yes) Then calling "make fate-vc1" will just return instead of throwing an error. Many (most?) of the test subsets converted so far behave in this way and it feels more logical and consistent to follow that line. We need to make a policy decision here that I can use as guideline for further conversions and (possibly) for updating existing tests. Diego _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
