Re: STk vs. Guile

From: Bernard URBAN <Bernard.URBAN_at_meteo.fr>
Date: Fri, 6 Sep 1996 10:39:40 +0200

>>>>> "Kevin" == Kevin K Lewis <lewikk_at_rockdal.aud.alcatel.com> writes:

    Kevin> Perry E. Metzger writes:
>> Anyone have any opinions comparing/contrasting STk with Guile?

    Kevin> Only some simple ones.

    Kevin> Guile, like STk, is supposed to provide a way to use Tk
    Kevin> with Scheme. Guile is also supposed to be an extension
    Kevin> language library.

STk can not natively be used as a library

    Kevin> So far, however, I have only heard comments about how Guile
    Kevin> should provide an object system ala STk (RMS said he
    Kevin> "definitely wants [an object interface to Tk] for Guile".
    Kevin> But there are no "plans" to do this. So I think this would
    Kevin> be like using STk, but without STklos (which is the biggest
    Kevin> benefit of STk, imo).

I personnaly ported Tiny-Clos to Guile-iii. With some hacks on the
*.stk files, it was possible to run simple STklos+Tk programs with
guile. But I still encounter problems with complex Text widgets.
I am not a Scheme specialist, so this shows it is a matter of effort
to bring STklos to guile.

Notice that currently, such a port is very slow, even when the
function call protocol is hard coded, as in STklos.

    Kevin> I think Guile would be best served by putting (something
    Kevin> very like) STklos on top of Guile.

I aggree ! STklos is very convenient.

    Kevin> Guile is based on SCM, a good-performance Scheme
    Kevin> interpreter. I can't comment on how fast it is compared to
    Kevin> that of STk, but I've been pleased with its performance in
    Kevin> the past. And SCM works well with SLIB, a nice Scheme
    Kevin> library from the same author. However, I believe STk works
    Kevin> well with SLIB, too.

Yes, it does. And my main argument to port STklos to guile was that
non graphical aspects are more efficient with guile.

B. Urban
Received on Fri Sep 06 1996 - 10:42:21 CEST