[MPlayer-dev-eng] make fatetest fails

[Reimar Döffinger]

On 6 Jul 2012, at 19:35, Ingo Brückl <ib at wupperonline.de> wrote:
> Reimar Döffinger wrote on Fri, 6 Jul 2012 18:44:12 +0200:
> 
>> On Fri, Jul 06, 2012 at 06:23:54PM +0200, Ingo Brückl wrote:
>>> I wrote on Fri, 06 Jul 2012 18:11:59 +0200:
>>> 
>>>> After downloading the samples for a ffmpeg fate test I though it might be
>>>> interesting to do such a test (the first time ever) for MPlayer as well.
>>> 
>>>> I've got lots of (i.e. 212) errors:
>>> 
>>>>  make: *** [tests/res/4xm/dracula.4xm.md5] Error 1
>>> 
>>> Well, I've just discovered that there is no .md5 file for this test.
>>> Couldn't we check before whether a test is reasonable?
> 
>> The idea from my side was that whoever runs into that case would just
>> copy the md5 that the test just created into tests/res and then commit
>> it.
> 
> Shouldn't at least two independently created identical md5 be necessary for
> that?
> 
>> Still, anything you can do to make it more usable is welcome,
> 
> I propose to skip at least samples we have no md5 for without causing an
> error for the moment (see patch).

Well, that makes it really painful to add the new ones.
Only generating the md5 without diffing seems nicer, however it might make it look like everything is ok if someone manages to run against an empty reference directory.