On IRC this week there was quite a bit of discussion/confusion about
the buffering system changes I was suggesting last week :-).

So this week I meandered in a slightly different direction from where I
had been going and implemented a second solution to the same problem of
buffering two files: I allowed a codec to do the work in its own way.
That work is at FS#10434 . Of course, the codec-specific buffering
function must be in the core (I don't want to much about with adding a
second entry point into the codec), but what's up there does work.

I'm going to spend most of next week drawing an illustrated guide to
how I envision the buffering system working in a backwards-compatible
way supporting multifile buffering. I'm pretty confident the
model I have works, but before anyone will accept it they need to
understand it. And people like pictures. Of course I'll also keep
polishing up the buffering-system patch I posted last week as well as
documenting it :-D.

Bryan Jacobs