[FFmpeg-devel] [PATCH 0/2]

[John-Paul Stewart]

Hi Ronald,

On 2021-11-28 9:02 a.m., Ronald S. Bultje wrote:
> Hi John-Paul,
> 
> On Sat, Nov 27, 2021 at 8:22 PM John-Paul Stewart <
> jpstewart at personalprojects.net> wrote:
> 
>> On 2021-11-27 8:51 a.m., Ronald S. Bultje wrote:
>>> Hi,
>>>
>>> On Fri, Nov 26, 2021 at 4:52 PM John-Paul Stewart <
>>> jpstewart at personalprojects.net> wrote:
>>>
>>>> There were no single channel examples and no 8 bit
>>>> examples.
>>>>
>>>> I have created four clips with SGI's original 'moviemaker' utility (so
>>>> they are guaranteed to be correctly formatted) to reverse engineer those
>>>> two pieces of metadata since the format is largely undocumented.  They
>>>> are under 1MB each and available at:
>>>>
>>>
>>> To prevent this from breaking in the future, it would be great if
>>> "as-small-as-possible" versions of these could be integrated into FATE
>> (our
>>> automated test system). How small can you make these so they still test
>> the
>>> bits-per-sample / channels features?
>>
>> I think they're already as small as I can practically make them.  The
>> biggest is only 600 KiloBytes.  In fact, my whole set of four is smaller
>> (2.0 MB) than the smallest existing test file (2.3 MB for 12345.mv) in
>> FATE's mv directory.
>>
>> In fact, the stereo16bit.movie example is redundant since there are
>> already samples with that combination.  So you'd really only need about
>> 1.4 MB of new test files.
>>
> 
> Alright. Would it be possible for you to submit a patch adding these other
> 3 samples to FATE then?

Frankly, it will be a while before I can submit a patch adding a test to
FATE.  It's beyond my current skill level.  I've started looking at the
documentation at https://trac.ffmpeg.org/wiki/FATE/AddingATest and some
of the existing FATE tests to use as examples.  But I've got a long way
to go before I can add a suitable test.