From: <al...@fo...> - 2003-02-07 19:39:46
|
On Fri, Feb 07, 2003 at 08:20:32PM +0100, Steve Lhomme wrote: > The idea of this system is to be able to help Hardware Players cache > data, regardless of the codec. If it's inside the container, the HP can > know in advance if it can read the file or not, regardless of the codec > (which hide how the data is cached). That means if more complicated > systems (than IBP frames) are used, the player can still know if it is > playable or not (without asking the codec). Well, keep in mind that this is generally not possible to do anyway without the codec (for most existing systems, and likely for future ones) because the data which is really important to cache is usually the _output_ of the codec's decompression process, not the input. A more useful way to do that sort of thing is probably to have the encoding codec provide general "requirements for decoding (max cache buffers needed, bitrate, etc)" information which can be stored as a guideline in the file headers. > It also allow to check if there are some bad references in a file, ie > damaged file. Shouldn't the container have other ways of determining if its file has been damaged (CRCs, etc)? -alex |