Re: Documentation with GCC3.1, (was: build error on solaris)

[Michael Livshin]

Neil Jerram <address@hidden> writes:

> [ should this move to guile-devel? ]

I CC'd there, at least.

> >>>>> "Martin" == Martin Grabmueller <address@hidden> writes:
> 
>     Martin> (1) We could change the build stuff to fire up Guile once,
>     Martin> and snarf all .x and .doc files in one run.  That would
>     Martin> probably break incremental building, though.
> 
> This is a possibility for the .doc files.  I don't think we need to
> involve Guile at all in the .x files; what we have already is fine for
> the .x's.

I'd even propose that we throw out the vile AWK stuff at all.  the C
preprocessor should be smart enough to allow us to dump the docstrings
(as they are) into the .x files, so that instead of looking like this:

  scm_make_gsubr (s_scm_make_vector ,   1 ,   1 ,   0 , (SCM (*)() )  
scm_make_vector );    

they would look like this:

  scm_set_primitive_doc_x (scm_make_gsubr (s_scm_make_vector ,   1 ,   1 ,   0 
, (SCM (*)() )  scm_make_vector ),
                 "Returns a newly allocated vector of @var{k} elements.  If a 
second\n"
                     "argument is given, then each element is initialized to 
@var{fill}.\n");

this frees us from worrying about AWK and whether Guile is already
installed or not -- the docstrings can be massaged by Guile itself
upon use.

this does leave the problem that the installed Guile might not have
the regexp feature, though.  I'm not sure this is a big problem in
practice -- who apart from embedded developers would install Guile
without regexps?

-- 
The software isn't finished until the last user is dead.