Re: @ref into @seealso

[Thorsten Meyer]

John W. Eaton wrote:
> So if you would prefer to handle this in a .m file, that
> would be good.  Either way, I think the transformations should be
> handled in one program rather than several.
>   
Actually, I think it should not all be done in one program:
As far as I understand, the techinfo command @ref does not only need the
name of the anchor it points to but also the name of the file, where it
can be found. That means, that whatever program expands the @seealso
commands has to know in which txi-file the individual function
descriptions are to be found. This implies scanning all txi files. The
present munge-texi program however only works on one txi file at a time
which is good because it means that the txi files can be remade
individually.

I propose to do the following:
 - add one octave program that extracts where the individual functions
are documented in the txi files and stores them in a file (like the
DOCSTRINGS files generated for the scripts and src directories)
 - rewrite munge-texi in octave (mainly because I do not know enough of
c++) and let it do the conversion of the txi files to texi files
including expansion of @seealso().

By the way: does the regexp function depend on an external library which
may not always be present or can I use it for the above?

regards

Thorsten Meyer