[FFmpeg-cvslog] r21598 - in trunk: ffmpeg.c tests/ref/lavf/ogg tests/seek.regression.ref

[Mike Melanson]

Mike Melanson wrote:
> M?ns Rullg?rd wrote:
>> Michael Niedermayer <michaelni at gmx.at> writes:
>>
>>> On Mon, Feb 01, 2010 at 11:44:54PM +0000, M?ns Rullg?rd wrote:
>>>> Michael Niedermayer <michaelni at gmx.at> writes:
>>>>
>>>>> On Mon, Feb 01, 2010 at 11:08:46PM +0100, michael wrote:
>>>>>> Author: michael
>>>>>> Date: Mon Feb  1 23:08:45 2010
>>>>>> New Revision: 21598
>>>>>>
>>>>>> Log:
>>>>>> Check pkt.pts against the recording time.
>>>>>> This fixes at least ogg encoding with -t where the file was 
>>>>>> slightly too long.
>>>>> yes i broke fate
>>>>> and fate doesnt tell me which tests faild :(((
>>>>> just 3 that at first glance could be correct
>>>> The stderr section should say.
>>> great ...
>>> we had 316 tests now just 127 are left ... if i knew which i could look
>>> into it ...
>>
>> Test 143/svq3 is being naughty.  It prints some non-utf8 characters in
>> the metadata, and python or sqlite chokes on that.
>>
>> My systems have a workaround[1] and are reporting all the other tests.
>>
>> [1] http://git.mansr.com/?p=fate;a=commitdiff;h=4d88f87
> 
> Updated mine too. Extremely disturbing that these characters make the 
> system choke. Makes me feel insecure.

Wait, this didn't fix the problem.

And that makes sense, too. In my investigations into the matter, the 
problem is not on the client-side; there's a problem with the PHP that 
receives the data and dumps it into the database.

-- 
     -Mike Melanson