Re: [ANNOUNCE] rep-gtk-0.18.4 (other scheme..)

["Chengqi(Lars) Song" <songcq gmail com>]

it seems there is a bunch of existing implementations of lisp with gtk binding. so is there anyway to use them directly instead of maintaining rep and rep-gtk, which could be a heavy work when gtk updates?

On Sun, 03 May 2009, Eli Barzilay wrote:

> On May  4, Chengqi(Lars) Song wrote:
> > 0) extra/guile-gtk
> >    Guile-gtk is glue code to make the GTK+ (the Gimp Toolkit)
> >    accessible from Guile
> > 1) clisp-gtk2: 
> >    ANSI Common Lisp interpreter, compiler and debugger (with gtk2
> >    enabled)
> > 2) clfswm: 
> >    A(nother) Common Lisp FullScreen Window Manager
> > 3) stumpwm-git: 
> >    A tiling, keyboard-driven window manager written in common lisp
> > 4) cmucl:
> >    CMU Common Lisp. There are two CMUCL bindings to the GTk+ widget
> >    set. Espen Johnsen's clg binds directly to the GTk event loop via
> >    the CMUCL FFI (GNU LGPL).  Gilbert Baumann's (older) binding uses
> >    a client-server model to avoid problems with mixing two event
> >    loops.
> 
> You should add plt scheme to the list -- it is full of useful
> libraries, has a good ffi, uses a jit to execute code relatively fast,
> it's very alive and under active developement, and it's big enough
> that I can't write any decent one-sentence feature list.  I'm one of
> the core developers, so I'll be happy to help with any needs -- like
> adapting glue code or module syntax.  (And in case anyone wonders,
> this is not a plug I'm dropping for nothing: I've been a sawfish user
> for a really long time, and one of the things that always bothered me
> was the choice of a language that is "somewhere between elisp and
> scheme".)
> 
> 
> > I think the advantage of guile is that it's used by gimp, [...]
> 
> No -- the gimp plan was to eventually move to guile, but this was
> never done (IIRC, the reason given was that it's too big or something
> like that).  So it's still using a very small (and not really good)
> Scheme interpreter.  FWIW, I think that sawfish is in a different
> place which justifies a good language: Gimp uses Scheme for writing
> automated scripts but it's a C application -- whereas sawfish is
> mostly written in Scheme (and with a good language, even more can move
> to Scheme, making it easier to hack).
> 
> This leads to another question that should be considered: is it better
> to have a C application that embeds a Scheme interpreter or is it
> better to go the other way and have a Scheme application with some
> library/ies that are written in Scheme.  I strongly believe that the
> latter is a more sensible choice -- and was done once with the scwm
> window manager (which I used before I switched to the more alive
> sawfish).
> 
> 
> > The advantage of clisp is that common lisp is a lot more powerful
> > than scheme. but im not sure if it's a good or bad thing for
> > developing sawfish.
> 
> I could say here that PLT Scheme is far more powerful than common
> lisp, but what I should really say is that any specific implementation
> of either Scheme or CL is far more powerful than that.  The real
> advantage of CL implementations is that more is standardize -- but for
> reasons that I can get to if anyone cares, that is not really a
> problem, and even less so for sawfish.  The disadvantage of most
> common lisps is that they tend to have a "monolithic repl" approach,
> which usually makes them less convenient to embed or to turn to a
> distribution.
> 
> The PLT Scheme point is that it is particularly powerful with a ton of
> useful stuff.  Of course this could be a problem: the installation is
> pretty big -- which is fine when you want to use it, but not fine when
> all you need is a window manager.  The solution for this in PLT is its
> ability to compile into "binaries" and/or distributions via its
> command line compiler tool.  A "binary" is really a copy of mzscheme
> (the plt scheme core) with byte-compiled scheme code attached; and a
> distribution is basically a file tree holding only the files that are
> needed for an application.
> 
> -- 
>           ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
>                   http://www.barzilay.org/                 Maze is Life!