Re: Elisp performance

[Ken Raeburn <raeburn@raeburn.org>]

On Aug 4, 2009, at 06:17, Andy Wingo wrote:
> Hello!
>
> (Was away for the weekend, but back hacking all week now.)

Welcome back!

> On Thu 30 Jul 2009 22:18, Neil Jerram <neil@ossau.uklinux.net> writes:
> Daniel Kraft <d@domob.eu> writes:
>>
>>> Lambda arguments are still always dynamically bound, which is  
>>> quite a
>>> pity as it inhibits tail-call optimization;
> Indeed, a pity. Though self-tail calls can be optimized... still,
> irritating.

In the compiler, or in the byte-code engine?

The Emacs byte compiler does not do self-tail call optimizations.   
Since the function value of a symbol can be copied out and/or replaced  
(and Emacs ships with the "advice" package which does both), the  
symbol's function value may wind up not being the function body being  
compiled; turning a self-tail call into a jump back to the beginning  
would cause observable behavior changes from what happens now.

On the other hand, I haven't looked at what elisp compiles to in  
Guile; maybe it's possible to have the byte-code engine detect when  
you're making a tail call to yourself (such as recursive calls when  
the function value slot hasn't been messed with) and optimize them on  
the fly?

Ken