Re: CVS script

[Maciej Stachowiak <mstachow@mit.edu>]

aarchiba@calum.csclub.uwaterloo.ca writes:
> Greg Badros <gjb@cs.washington.edu> writes:
> 
> > And, not to beat a dying horse, *documentation*!  I can't begin to
> > describe how much better I am at using scwm now that Maciej, Sam, and I
> > have worked together to create an interactive documentation system.  Sam 
> > wrote a nice Emacs interface to the documentation that is extracted from 
> > our source code, and all the functions are as easy to learn about as
> > Emacs lisp documentation.  It's a joy to use, and would be even better
> > if we linked to the R5RS info pages on scheme, and if Guile procedures
> > were documented in the same way.  Scwm and its scwm-mode for Emacs is a
> > great way to learn about Guile and could easily evolve into a great
> > scheme development environment.
> > 
> > Greg
> 
> Is there any reason not to apply the current scwm doc engine to guile?
> (Apart from requiring time and effort)  Is there any likelihood of
> having to rewrite the in-code doc strings if guie were to switch over
> now?
>

The doc string format we have is not likely to change (and works
well), but there are a few changes that I think need to be made to the
doc extractor before it should be shipped with guile:

* Rewrite in guile (current version is perl).

* Figure out a way to have documentation get disassociated from a
primitive if you re-define it as something else. Basically there needs
to be a procedure property giving the file to look in (there already
is one for name) and these need to be checked by the new documentation
procedure rather than just the symbol. Basically, if I (define + -) I
should get the - docs when I do (documentation +).

* Find a nice way to integrate docstring extraction with the build
system.

* Figure out pre-processor issues mentioned by Jim (Guile actually
provides some primitives conditionally, so this is relevant).

* Make it work with primitives declared using the normal Guile
SCM_PROC macro rather than the slightly less redundant SCWM_PROC that
Greg made all the scwm code use.

Jim may want other issues addressed as well.

 - Maciej