[FFmpeg-devel] [PATCH 3/3] configure: fail if autodetect-libraries are requested but not found
andreas.cadhalpun at googlemail.com
Fri Dec 2 01:44:41 EET 2016
On 01.12.2016 01:34, Michael Niedermayer wrote:
> On Thu, Dec 01, 2016 at 12:22:50AM +0100, Andreas Cadhalpun wrote:
>> @@ -6381,6 +6410,11 @@ for thread in $THREADS_LIST; do
>> +# Check if requested libraries were found.
>> +for lib in $AUTODETECT_LIBS; do
>> + requested $lib && ! enabled $lib && die "ERROR: $lib requested but not found";
> This must be after check_deps as that can disable vaapi
I agree, but...
> unless i miss something
...it is already after check_deps. ;)
> also in the same light i think things like:
> enabled vaapi &&
> check_lib va/va.h vaInitialize -lva ||
> disable vaapi
> should disable vaapi through having a entry in vaapi_deps= ...
There is already va_va_h in vaapi_deps and I'm not sure, what else
you'd like to have there. However, it seems independent of this patch.
> And then check_deps should check for requested when disabling things
> that way it should be possible to print why something is being disabled
> If that works that extra information should be quite usefull
More information never hurts, so I added some specific error messages
> also off topic but this can eventually be extended to cover more
> things than just these libs
> actually i wanted to cleanup configure and that stuff but
> its one of these things that kept sliding down on my todo ..
That's probably because as long as configure works a cleanup
isn't really important...
On 01.12.2016 09:48, Carl Eugen Hoyos wrote:
> Needs a Changelog
Added, updated patch is attached.
> and a NEWS entry.
As in "put the Changelog entry on the website"?
Fine for me if you think that'd be useful.
> Have you tested the patch?
Of course, and I made sure that the newly added functionality actually works,
however I can't possibly test all combinations of configure options in all
sorts of environments for regressions, so help in testing e.g. the more
exotic ones would be welcome.
> It was tried before and the result were different regressions iirc.
Can you provide links to previous attempts and their regressions so that
I can make sure they don't happen with my patch?
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 4594 bytes
Desc: not available
More information about the ffmpeg-devel