From: Robin K. <kom...@my...> - 2002-11-24 19:29:27
|
Mike Melanson wrote: > On Sun, 24 Nov 2002, Guenter Bartsch wrote: > > > ...of course you could also have a look at the input_stdin_fifo.c sources and add a > > few printfs to find out what's going on and maybe commit a bugfix thereafter ;> > > Oh, so you mean it is not supposed to do that? I will probably > have to plunge into the source anyway to figure out why it does not seem > to seek forward the proper number of bytes either. Is forward seeking a capability of input plugins that I can always depend on existing regardless of whether INPUT_CAP_SEEKABLE is set? Would be nice for me to get rid of the skip_bytes routine in demux_wve if it is. A unified system for back seeking similar to the mark() and reset() methods of java.io.InputStream would also be, IMHO, a good idea. I would add a mark() function for setting a mark and simply allow back seeking up to the mark. There would also need to be a clear() function for removing the mark. This would also dispense with the need for preview buffers and be far more flexible. It would simplify some the problems of data structure traversal that the lack of back seeking brings. For example, reading an arbitrarily long null terminated string. -- Wishing you good fortune, --Robin Kay-- (komadori) |