[FFmpeg-devel] [PATCH] Reduce exessive seeking in mov demuxer

Baptiste Coudurier baptiste.coudurier
Fri Oct 12 13:41:35 CEST 2007


Hi

Michael Niedermayer wrote:
> Hi
> 
> On Sat, Oct 06, 2007 at 12:40:51AM +0200, elupus wrote:
>>> Yes, I totally agree.
>>> Im still thinking about a nice way to handle piped or streamed files
>>> where it might be better to read by position.
>>> Demuxing that way for url_is_streamed files might be ok.
>>>
>>> What do you think about that ?
>>>
>> In general that sounds like a good plan, but there is a problem with that 
>> for http streams where seeking actually is allowed, just not fast. Setting 
>> url_is_streamed then isn't a good approach as it breaks other files + 
>> changes how the mov header is read.
>>
>> I'm quite sure the supplied patch will be accepted as is. if nothing else 
>> cause it does mix in some cosmetics + code dup. But maybe something like it 
>> would.
>>
>> It can force up to a second of buffered data wich might be unacceptable. 
>> But for my purpose it'll be fine (we buffer almost 8 seconds :/).
>>
>> What do you guys think?
> 
> the patch is trashed that is linewrapped ..
> also it should be split in the url_is_streamed and the 1 second part
> the first i think we all agree is good

Yes, though it won't enable "cat file.mov | ffmpeg -i -" "as is", so it
is not applicable. I have one question pending to enable that, related
to streamed seek.

> iam not so sure about the 1sec part, but iam not objecting against that
> if baptiste thinks its usefull

It may be useful, one user reported very slow demuxing through http due
to excessive seeking, and patch seems to fix the problem.

-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
SMARTJOG S.A.                                    http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
Phone: +33 1 49966312




More information about the ffmpeg-devel mailing list