On Tue, Mar 23, 2010 at 12:01:55PM -0700, Baptiste Coudurier wrote:
On 03/23/2010 11:59 AM, Michael Niedermayer wrote:
On Tue, Mar 23, 2010 at 06:54:23PM +0000, M?ns Rullg?rd wrote:
Michael Niedermayer<michaelni at gmx.at> writes:
On Tue, Mar 23, 2010 at 10:57:04AM -0700, Jason Garrett-Glaser wrote:
On Tue, Mar 23, 2010 at 10:21 AM, Michael Niedermayer<michaelni at gmx.at> wrote:
On Tue, Mar 23, 2010 at 01:59:46PM +0200, Kvikant, Christian wrote: > M?ns Rullg?rd wrote: >> Sorry, but what is the point of this? There is already a fantastic, >> open source H.264 encoder in x264. > I agree totally. FFH264 is not up to par nether regarding > quality nor size. The only reason for this exercise was > licensing. If the feeling is that FFH264 will never become part > of FFmpeg, then we will keep it as a separate patch.
a simple and small encoder could be usefull for regression testing.
If the encoder uses practically no features it's totally useless for regression testing...
It shouldnt be hard to use lots of features and just write random values ;) for regression tests that would be fine.
Then we are no longer talking about an encoder in the usual sense. Besides, ensuring the maximally awkward values are used is _hard_. The conformance tests already do that, so there's no need to duplicate that work.
testing the conformance streams need a lot of time, its not something i do before every commit. but make test is something i try to do before commit unless its a trivial change with that workflow in mind, a small and fast h264 test would be usefull
Testing the muxers code path with h264 is also useful.
that reminds me IIRC that tests/copycooker.sh is not run by make test yet [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB I am the wisest man alive, for I know one thing, and that is that I know nothing. -- Socrates