[plt-scheme] A Couple of Questions on DrScheme/Mzscheme

From: Mike Lin (mikelin at MIT.EDU)
Date: Thu Sep 26 10:38:39 EDT 2002

I just want to raise the point that it is rare for large, "real" software
applications to be written entirely in one language or environment. It is
perfectly reasonable and not uncommon to find huge mission-critical
applications with, for example, a Lisp backend, VB UI and Tcl scripting
interface. Besides allowing us to use the best tool at hand for the job, I
think it is arguable that this heterogeneity forces us to observe certain
abstraction barriers and modularity constraints among the various components
of the application.

Not that it doesn't cause big maintenance and design headaches, of course;
but gluing together heterogenous components is part of what software
engineering is all about, and I don't think it's at all clear that we would
on the whole be better off if we all used one programming system.

So I would argue that it is less important for us to have full-featured GTK
bindings than to ensure that we have full-featured and easily accessible
bindings with other programming systems. For example, I was recently
introduced to a little script that generates JNI classes from OCaml modules.
This kind of automation eases the (I consider) inevitable necessity to glue
together components written in different environments; the value of this
automation, if you have ever used Visual Studio .NET, is something that
Microsoft understands very, very well.


----- Original Message -----
From: "Michael Brickenstein" <bricken at rhrk.uni-kl.de>
To: "Noel Welsh" <noelwelsh at yahoo.com>
Cc: <plt-scheme at qua.cs.brown.edu>
Sent: Thursday, September 26, 2002 9:56 AM
Subject: Re: [plt-scheme] A Couple of Questions on DrScheme/Mzscheme

>   For list-related administrative tasks:
>   http://list.cs.brown.edu/mailman/listinfo/plt-scheme
> Hello!
> > Some thoughts on prettier GUIs:
> I think this all is definitively a big problem. We really need a better
> in particular a better canvas. Drawings look badly (no antialiased lines),
> and it should be object oriented like evas or gnome-canvas with features
> "Give me the objects at this points".
> > - GTK could be bound using SWIG (comments from those
> > with more experience with SWIG?)
> >
> Gtk maybe, but we also need a good canvas (guile's support for
> exists more or less only on paper).
> Gnome-canvas is harder, not so easy to make with just SWIG I think.
> Variable numbers and types of arguments, ugly.
> I've recently done some major work with swig for a scheme binding of evas
> canvas, which will be used in enlightenment 17).
> Such a think needs always handwritten code for critical points like
> Swig is however not very mature for using with mzscheme. But I've got some
> little example working and have some fun.
> I'll let you know in this list, when I publically release my code.
> Michael

Posted on the users mailing list.