[FFmpeg-devel] [PATCH 2/4] Add replay_gain and find_peak_sample options to libmp3lame

[Giovanni Motta]

I've re-read the description of #3577 and I am not even sure we are talking
about the same "Lame tag" here.

What this patch set adds is a Lame info tag extension to the Xing header
described here:
http://gabriel.mp3-tech.org/mp3infotag.html (see "*Suggested Info Tag
extension fields + layout* ").

Some of the encoder settings described in #3777 are stored here, but the
main point of this extension is to support gapless playback (delay and
padding) and gain control (replay gain and peak sample).
G.


On Wed, May 28, 2014 at 8:16 AM, Giovanni Motta <giovannimotta at google.com>
wrote:

> Patch 1/4 should amend the texi file for a typo (two extra dashes), so 1/4
> and 2/4 are parts of the same patch.
>
> > And I suspect not all four patches fix 3577 but one
> > (or maybe more than one but that is already unexpected.)
>
> Define "fixes" :-)
>
> 3577 is more of a feature than a bug (ffmpeg does not create and add the
> lame tag to the Xing header). I could remove the statement from 1/4-3/4,
> but 1/4-3/4 are pointless without 4/4 (some settings for the tag are added
> [1/4-2/4], the tag is created and copied to extradata [3/4], but not added
> to the header until 4/4). So, is 4/4 "fixing" the 3577? Any better way of
> specifying dependency?
>
> > I cannot judge how this should (or even can) be split correctly.
>
> I think we agree here :-)
>
> Original submission, was a single patch and I was told to split the new
> options added, and then separate libavcodec and libavformat changes, this
> is why the three patches (well, plus the amend on the doc file). I
> understand the logic, but I am not sure how you want me to handle this case.
>
> Any suggestion?
>
> Thanks
>
> G.
>
>
>
>
>
>
> On May 28, 2014 2:04 AM, "Carl Eugen Hoyos" <cehoyos at ag.or.at> wrote:
>
>> Giovanni Motta <giovanni.motta <at> gmail.com> writes:
>>
>> > Fixes ticket #3577
>>
>> Something is definitely wrong with your patchset;-(
>>
>> This patch (2/4) says it adds find_peak_sample but
>> it was already added in 1/4...
>>
>> And I suspect not all four patches fix 3577 but one
>> (or maybe more than one but that is already unexpected.)
>>
>> Sorry, I cannot judge how this should (or even can) be
>> split correctly.
>>
>> Carl Eugen
>>
>> _______________________________________________
>> ffmpeg-devel mailing list
>> ffmpeg-devel at ffmpeg.org
>> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>
>