Re: Faster lisp engine
On Fri, Aug 14, 1998 at 10:26:44AM +0200, Michael Sperber [Mr. Preprocessor] wrote:
I guess this was my entry for the obscure comment of the day: What I
meant was: Strong typing doesn't buy you much in way of performance.
I think Martin was right in the analysis that Elisp isn't the
bottleneck. However, a faster substrate would allow us to move
*a lot* of stuff currently written in C into the substrate.
The one single thing which would really make substrate performance rip
would be a native-code compiler.
Strong typing (or more precisely type inference) can, I think, make a
hell of a difference when you're generating native code. In
particular if you allow inline generation of lrecord member accessors
for simple types.
An elisp frontend for gcc anybody ? :-)
OG.