[FFmpeg-devel] [PATCH 2/5] movenc: simplify codec_tag lookup

[Paul B Mahol]

On 7/3/17, Marton Balint <cus at passwd.hu> wrote:
>
>
> On Mon, 3 Jul 2017, Hendrik Leppkes wrote:
>
>> On Mon, Jul 3, 2017 at 5:39 PM, wm4 <nfxjfg at googlemail.com> wrote:
>>> On Mon, 3 Jul 2017 16:17:42 +0100
>>> Derek Buitenhuis <derek.buitenhuis at gmail.com> wrote:
>>>
>>>> On 7/3/2017 2:18 AM, Michael Niedermayer wrote:
>>>> > breaks fate
>>>>
>>>> I'll look into it tonight; busy today.
>>>>
>>>> .
>>>> .
>>>> .
>>>>
>>>> Aside:
>>>>
>>>> I'll just add, though, that these two word 'breaks fate' emails
>>>> are kind of obnoxious when the test in question was added days
>>>> after I sent the set, so I couldn't have possibly tested against
>>>> it, and the commit that added the test and this email has /zero/
>>>> info about what the test actually tests (a bug id is not a commit
>>>> message).
>>>
>>> This. These opaque fate tests are so much work to get around. It went
>>> far enough that I added bullshit to ffmpeg.c to get around some of the
>>> questionable tests.
>>>
>>> Also, TRAC issue numbers have 0 information contents. Even if you go
>>> through the obnoxious process of looking it up on TRAC and trying to
>>> extract iunformation from a confusing discussion with a confused user
>>> (99% of TRAC issues), TRAC could easily go away. It already happened
>>> once, and some of the bug numbers in old commits obviously don't match
>>> with what's on current TRAC.
>>>
>>
>> I agree, this test could've easily been named something useful, like
>> fate-mp4-copy-eac3 or whatever namespaces we use for mp4 tests, which
>> would convey the same information without having to lookup the ticket
>> on trac.
>>
>
> Can't the project pay someone to make fate tests from fixed trac tickets?
> Or make this an outreachy goal, like API tests, or something like that.

What would then remain for Carl and folks?