[FFmpeg-devel] [RFC] swscale regression tests

[Måns Rullgård]

Ramiro Polla <ramiro.polla at gmail.com> writes:

> On Sat, Jul 31, 2010 at 6:15 PM, Ramiro Polla <ramiro.polla at gmail.com> wrote:
>> 2010/7/21 M?ns Rullg?rd <mans at mansr.com>:
>>> Ramiro Polla <ramiro.polla at gmail.com> writes:
>>>> There are 42 input/output pixel formats, 5 algorithms being tested,
>>>> and all combinations of (64,96,128)->(64,96,128) conversions. That's
>>>> 79380 tests. Ideas to rework the tests are welcome.
>>>
>>> OK, then splitting it up probably isn't practical. ?How about one
>>> checksum per test? ?Did someone already suggest that?
>>
>> I'm working on this in the other thread to add a switch for checksums
>> to swscale-test. Do you think splitting the tests in input formats is
>> good enough, or do you prefer another approach like Vitor?
>
> Now there's support on swscale-test to specify input/output formats,
> and also to specify a reference file to avoid unnecessary SSDs. This
> halved the time it takes to run the full test (when a reference is
> given), so it should be fast enough to run regtests.
>
> OTOH with the addition of crc output, the output is 7.something mb and
> the bz2'd output is 500 something kb. It would be nice to have the
> results for many different optimizations too, so that would be quite a
> lot of data. Does anyone have any good idea on where that should be
> made available for the tests to run? (FFmpeg svn, some other repo,
> fate-suite/, ...?)

Is the output highly redundant in some way?  Perhaps it could be made
less verbose.

-- 
M?ns Rullg?rd
mans at mansr.com