Re: Package Scope Recommendations for File_Ogg

From: Date: Thu, 14 Aug 2003 08:30:52 +0000
Subject: Re: Package Scope Recommendations for File_Ogg
Groups: php.pear.dev 
Request: Send a blank email to pear-dev+get-19715@lists.php.net to get a copy of this message
Hi Stefan, > didn't intend to cause you so much headache - sorry for that :-) On the contrary, it's been a valuable experience. :) > Well I didn't really think about the theoretically possible point > combined with several parallel streams and that one can end before > the other and a new might start and ... Okay okay, this seems a bit > out of scope. Yes, it's a bit of a headache to implement. Glad to hear you opinions on the scope of the project. > Suggestion: How about limiting it to parallel streams? This way it > would be possible to detect Theora-video with vorbis-audio or similar > correctly. Parallel streams WILL be supported, featuring any of the four current Ogg projects (Vorbis, Theora, Speex and Flac). > What files do you think exist "in the wild life"? Okay I could think > of an ogg-file that contains some audiostreams following each other > ... maybe if they switch bitrate inbetween or so. But I don't think > this occurs very often. And if you have a look at mp3, mp3-players > etc. you will see (as one example) that most programs don't handle > intermediate bitrate-changes (if its not marked as vbr) correctly. It > too hard to detect things like this correctly, to parse everything > (if its a huge file) etc. I intend to have a chat with some of the Ogg developers (if I can find them on an IRC channel) to see if they can point out "unusual" physical streams, such as a mixture of Vorbis and Speex, for example. Personally I can't really see the point of concurrant streams unless it forms part of a audio or video stream. > So from my point of view it would be enough to concentrate on having > parallel streams (video+audio, or maybe 2x audio, maybe some > additional codec for hell-knows-what in future development). OK. So it will be. > Using libvorbis etc. directly via a pecl-interface, as proposed by > Patrick, doesn't seem a good solution for me. Okay, it would save you > reinventing the wheel. But if it's possible to solve this problem in > "clean" php (and without taking too much amount of memory and time > for parsing) I would favor a php-pear- and not a pecl-class. With a > pecl-solution you are dependent on libvorbis etc. I've pointed out that both solutions could be offered. Since I've no experience in C, I suggest someone else might like to take a look at it. > PS: Keep up your motivation, David! Thanks! Just having fun at the moment. :) Regards, David

« previous php.pear.dev (#19715) next »