[FFmpeg-devel] [PATCH 2/2] fate: use do_md5sum instead of the md5 protocol for md5 fate tests

James Almer jamrial at gmail.com
Sat Jun 17 03:41:54 EEST 2017


On 6/16/2017 9:36 PM, Marton Balint wrote:
> 
> On Fri, 16 Jun 2017, James Almer wrote:
> 
>> On 6/16/2017 8:24 PM, Marton Balint wrote:
>>> The md5 protocol has no seek support, but some tests use seeks. This
>>> changes
>>> the fate tests to actually create the output files and calculate the
>>> md5 on the
>>> written files, which also makes the tests independent of the size of
>>> the output
>>> buffers and output buffering in general.
>>
>> The fact md5() forced non seekable output came in handy to test certain
>> codepaths, especially in matroska. Perhaps you could instead add a new
>> fate-run.sh function for this change and port those tests that use
>> output formats that don't behave any different in non seekable output?
>>
> 
> IMHO when most people are using the md5 test, they are typically not
> aware the lack of seek support. Considering how many md5 tests are used
> all over fate, I'd rather create a test named md5pipe which uses the md5
> protocol (this name better reflects the streaming property of the test
> as well), and change the matroska tests to use that, if you really want
> to test the streaming output.

Yes, that also works. The idea is to keep a working fate function that
can easily let us test muxers when using non seekable output.

> 
> BTW in the matroska muxer, should a non-seekable output and using the
> -live option produce the same result? Because there are some places in
> the code which only checks if -live is set, and I am not sure if that is
> intentional.

Probably not. All the code related to the -live option is nowadays
somwhat precarious after further development and merges took place in
the muxer.

> 
> Regards,
> Marton
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel



More information about the ffmpeg-devel mailing list