Re: Why is Elisp slow?

[tomas]

On Thu, May 02, 2019 at 09:13:28PM +0200, Marcin Borkowski wrote:

[...]

> > (My personal rant to package authors is that they are way too eager to
> > implement stuff in Lisp which cannot possibly be fast enough or
> > scalable enough, instead of urging Emacs to provide new C primitives
> > and core features, or, better, submitting patches to implement such
> > features in C.  Two cases in point are linum-mode and fci-mode.  End
> > of rant.)
> 
> That is an interesting POV, although I'm not sure if I agree.  IOW,
> I fully understand why people might do so.  I, for one, consider myself
> pretty fluent in Elisp and hardly literate in C.

This approach is known as "Ousterhout's dichotomy" [1], after John Ousterhout,
author of Tcl (which was conceived as a very simple scripting language
meant to extend existing applications). Tcl grew later a bytecode
interpreter (as did elisp) -- making the C interface significantly
more complex (but the interpreter significantly faster).

Walking the trade-off between simple and efficient is interesting.

Cheers

[1] https://en.wikipedia.org/wiki/Ousterhout%27s_dichotomy

-- tomás

signature.asc
Description: Digital signature