Re: Shrinking the C core

[Po Lu <luangruo@yahoo.com>]

Arthur Miller <arthur.miller@live.com> writes:

> Very interesting discussion going on for a long time.
>
> I think you are all correct, and wrong to an extent, but I believe that
> nobody has touched the fundamental issue: emacs design is flawed beyond
> repair for todays machines. Not necessarily in pejorative meaning, but
> to repair Emacs you would have to significantly rework internals, to the
> point of entire design rewrite. Emacs is a child of its time (like
> everything else). It was designed for the time of single-core slow
> machine, and its design makes sense in that perspective. However, for
> todays multicore machines, the fact that a lisp machine is slapped on
> top of an existing text editor (Gosslings I guess), and everything is
> shared via global state, can't be addressed in any other way but to
> rewrite Emacs core from ground up. No amount of patch slapping onto the
> current design can compensate for the lack of appropriate desing.

Unix was designed for 16-bit uniprocessor machines, where the only form
of ``interlocking'' was:

   int x = spltty ();
   splx (x);

But today, both free BSD Unix and proprietary Unix scale to SMPs with
hundereds of processors, exploiting intricate interlocking around
individual kernel data structures.  The perfect antithesis to your
standpoint...