[FFmpeg-cvslog] r20839 - trunk/libavformat/id3v2.c

Baptiste Coudurier baptiste.coudurier
Mon Dec 14 04:19:14 CET 2009


Hey guys,

On 12/13/09 5:53 PM, Michael Niedermayer wrote:
> On Sun, Dec 13, 2009 at 05:35:57PM -0500, Alex Converse wrote:
>> On Sun, Dec 13, 2009 at 4:02 PM, michael<subversion at mplayerhq.hu>  wrote:
>>> Author: michael
>>> Date: Sun Dec 13 22:02:43 2009
>>> New Revision: 20839
>>>
>>> Log:
>>> More entries for ff_id3v2_metadata_conv from ffmbc.
>>>
>>
>> Most of these changes being pulled from FFmbc seem non controversial.
>> Is there a reason they weren't sent upstream first? Hoarding patches
>> helps nobody.
>
> I was thinking the same, besides i really wish i had a patchset or
> repo with history from where i could extract this stuff., i just have a
> tar.gz.
> And there are some changes that i dont know why they are there so i dont
> know what to do with them.

First, I'm glad that some changes were interesting and applied.

Also I'd delighted to have a place to push the tree, so people can have 
a look and cherrypick the good patches.
I believe it could be an efficient and easy way to work, especially for 
developers.

> [...]
>
> the rest is merged, wrong or bikeshed. The above is mostly stuff that
> still needs to be looked at and or understood better.

Some modifications are experimental. Especially the pts handling in 
ffmpeg.c and the removal of pts/dts interpolation for formats that have 
this information. For example mkv dts computation is not working 
properly with h264. For stream copy, computing codec delay from avcC 
and/or decoder and interpolating dts on the muxer side is working better.

The average frame rate code is working better for formats having the 
information stored: mkv and mov in particular.

The h264 parser can parse iso media format although it's not that 
useful. The parser also output merged fields which makes AVCHD stream 
copy working.

If you have any question, I'll be glad to explain the changes.

Also all the changes to files I maintain will be applied of course, 
provided that the changes are not specific to my usage and acceptable :)

-- 
Baptiste COUDURIER
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
FFmpeg maintainer                                  http://www.ffmpeg.org



More information about the ffmpeg-cvslog mailing list