Re: [Fsedu-developers] Proposal for course: Scheme

[Peter Minten]

Stephen Compall wrote:
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Tuesday 01 April 2003 10:54 am, Peter Minten wrote:
> > I will use Texinfo in the book, mainly to try how good it is in
> > writing books. The Scheme dialect used is, of course, Guile. The book
> > will consist of two parts: theory and practice. The theory covers
> > roughly the following topics:
> > * Data types (including conses and vectors)
> > * Functions
> > * Control functions
> > * Structures
> > * Objects
> 
> There are two ways to go about it, I think: teach theory, then
> practice; or just do both at the same time.  Far be it for me to
> decide which is better; I've seen bad implementations of both
> strategies.

Well, how about this, we use the practice stuff in the theory examples where
possible. I mean explaining how to use a function by drawing a line on the
SchemeCanvas.

> > The practice part consists of writing scheme scripts for two
> > programs. The first program is a small GTK canvas window which has a
> > scheme interface, the second program is GNU Robots.
> 
> Yes, the second one is especially cool.
> 
> > Two competition will be attached to the practice, one for each
> > program. The SchemeCanvas competition is about who writes the coolest
> > scheme drawing program and the Robots competition about who writes
> > the best robot.
> 
> All depends on what you mean by "best" ;)

Yeah. Well, for robots it the best scoring robot. For SchemeCanvas the coolest
drawing, though that's more of a jury matter.
 
> > Note that the book will not cover writing Guile interfaces to C code,
> > that's something for another book.
> 
> If Guile users aren't comfortable with adding subrs to their modules
> and programs, then they might just switch to pure C/C++ programs just
> to be able to use some random library.  So I think it should be
> included as a section..."skip this if you don't know C"

OK, but if we're going to include it, it should be worked out decently, the scm_
interface is quite big and it needs a lot of explanation.

> I'm speaking here as an advocate of replacing most C programming with
> higher-level languages, and only interfacing into C for the stuff that
> has speed problems.  I don't know if fsedu is the place to advance
> this agenda, but anyway....

I'm personally no fan of using C as much as currently the trend and I believe
it's a good idea not to commit to this trend in FSEDU. That meX-Mozilla-Status: 
0009languages where the language used does not really matter. To make this
work we should have a preffered-language policy. I suggest scheme here, since
it's high level, GNU standard, and allows easy interfacing to C code when
needed.

Greetings,

Peter