On Wed, Jun 29, 2011 at 01:24:46PM -0400, Justin Ruggles wrote:
> On 06/29/2011 11:06 AM, Måns Rullgård wrote:
> 
> > Diego Biurrun <[email protected]> writes:
> > 
> >> On Wed, Jun 29, 2011 at 03:42:15PM +0100, Måns Rullgård wrote:
> >>> Diego Biurrun <[email protected]> writes:
> >>>
> >>>> --- a/Makefile
> >>>> +++ b/Makefile
> >>>> @@ -185,7 +185,7 @@ distclean::
> >>>>  config:
> >>>>          $(SRC_PATH)/configure $(value LIBAV_CONFIGURATION)
> >>>>
> >>>> -check: checkheaders
> >>>> +check: alltools checkheaders examples testprogs
> >>>
> >>> Why?
> >>
> >> This is supposed to be a convenient target that developers can run to
> >> make sure they have not broken compilation of targets besides "all".
> >>
> >> It might either include fate or be made part of fate (eventually).
> > 
> > The various test programs should be compiled and *run* as individual
> > fate tests, not merely built in bulk like this.
> 
> +1. Fate basically replaced these built-in tests as the primary test
> system without taking most of them into account. They tend to be
> unmaintained, and having them in Fate would give a big red flag (or
> yellow flag) that something is wrong.
> 
> It also might be an incentive to improve some of them (e.g. why do we
> have snow and h264 tests? we should instead have separate tests for dwt,
> exp golomb, etc..).

This is correct, but I consider my patch orthogonal to converting (some
of) the tests into FATE targets.  The tools and examples should also be
kept compiling and they are unrelated to FATE.  Maybe I could add fate
and all as dependencies to the check target, then running 'make check'
would compile everything and run all tests together.

Diego
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to