From: owner-xemacs-beta(a)xemacs.org [mailto:owner-xemacs-beta＠xemacs.org]On Behalf Of Michael Sperber [Mr. Preprocessor] Sent: Friday, July 17, 1998 12:29 AM To: xemacs-beta(a)xemacs.org Subject: Re: scheme - my opinion >>>>> "RP" == Reginald Perry <perry(a)zso.dec.com&gt; writes: RP> The problem with scheme is that while its a great abstraction RP> building tool, the baseline may be a bit too low. I remember RP> reading that Jeffrey Siskind moved to scheme and it took him five RP> years to get his abstractions working to his liking. I just looked at his code again. Most of is application-specific abstractions geared towards vision and language processing software. Most of it is, therefore, not in CL either. RP> OK. I may lean more toward scheme. Here is what would make for a kick ass RP> system. All IMHO of course. Once it comes to the crunch, we'll have no disagreement whatsoever. We're already on the same wavelength, except you're on top, and I'm on bottom. :-)

RP> A module system like SML; Higher order modules included. Scheme 48 and MzScheme have this currently, but the system in Scheme 48 is due to be replaced by something better soon.

RP> CLOS like object system + a MOP like STk. With this you really RP> don't need structures and you can use the MOP to properly expose RP> the object-based C code in the lower layers. Is there any Scheme object system you like? The ones in MzScheme and RScheme come to mind, RScheme being a scaled-down CLOS and MzScheme being a scaled-up Java object system.

RP> A condition system with conditions being objects like Common Lisp. Yes. RP> Generational garbage collection with excess space being given back RP> to the OS after a long period of non-use. I like to run my emacs RP> for months on end without restarting it. :-) Absolutely. I have the generic interface implemented in Scheme 48 and RScheme's collector hooked up to it. Next step is to sponsor a Master's student to implement the same interface in XEmacs and hack a suitable collector. RP> A fast scheme engine. This is the tough one. Scheme48 is currently a bit too RP> slow for this. Of course clisp has a lot of hand coded C/assembly so maybe RP> making a better prescheme compiler would do the trick. Huh? Last I checked Scheme 48 was about as fast as CLisp. (Note, BTW, that its bytecode interpreter is in Scheme---very little of the system is in C.)

-- Cheers =8-} Chipsy Friede, Völkerverständigung und überhaupt blabla