[gst-devel] 0.9 proposals

[Owen Fraser-Green]

Hi,

On Sat, 2004-12-04 at 14:22 +0100, Ronald S. Bultje wrote:
> Given that I've written some demuxers, let me please comment on this.
> First of all, I'm mostly assuming bytestream here. We've got some other
> ways, but all demuxers but ogg use bytestream. 
> *snip*
> So basically, we're pretty screwed here. Writing chain-based demuxers
> (or state machines, as Dave calls them) is not all that good of an
> option.

I may be wrong but I assumed Dave was thinking of keeping them as loop-
based elements but which ask for chunks at the beginning of their loop
functions and then chew on the chunks. This is of course close to being
chain-based except that you can ask for chunks which correspond to the
segment you're reading instead of having to join bits of buffers
together when a segment spans two or more buffers. Then the it's easy to
make the code requesting the segment handle interrupts etc. as opposed
to the normal was of requesting field-by-field.

It strikes me amid all of this that maybe the bytestream interface is a
bit broken. It's pretty irritating both to code and to read to have to
check for, and handle, interrupts for every field the plugin wants to
read. Would it be possible to implement bytestream so that it calls a
callback in the plugin when an interrupt occurs, then the
bytestream_read (or whatever the call is) only returns upon success (or
complete failure) effectively suspending the state of the plugin in the
middle of its data chewing until the event is dealt with?

Cheers,
Owen