Re: Patch for fields of `struct buffer'

[Tom Tromey]

RMS> Thus, it is a mistake to do much work now, or complicate the code now,
RMS> or cause a slowdown now, for the sake of preemptive thread switching.

Stefan> AFAICT, none of what he did is done because of preemptive threading.
Stefan> IOW, I don't think you can do it much more simply, even if you assume
Stefan> a cooperative threading model.

Yeah, you could.  On a thread-switch, you could unwind all the old
thread's bindings, then restore the new thread's bindings.

This would let all the existing C code work without indirections or
macro trickery; the downsides are that you have to have user-space
control over thread switching, and thread switching becomes expensive.

Tom