Re: [Texmacs-dev] literate programming for TeXmacs

[Felix Breuer]

Hello again,

and sorry for the long silence. Now:

On Tue, 24 Jan 2006 15:38:29 +0100
David MENTRE <address@hidden> wrote:

> Hello Joris and Felix,
> 
> 2006/1/24, Joris van der Hoeven <address@hidden>:
> > On Mon, Jan 23, 2006 at 04:52:34AM +0000, Felix Breuer wrote:
> > > To avoid misunderstandings: I think there were actually three concepts
> > > in discussion, namely
> > >
> > > 1) David's original idea: Use TeXmacs to work on source files that have
> > >    tm markup as comments.
> > > 2) My original idea: Generate (multiple) source files (without comments)
> > >    from a single TeXmacs document.
> > > 3) What I am currently suggesting: Serialize a TeXmacs document as a
> > >    source file where TeXmacs markup is commented. (This is a variant
> > >    of David's suggestion.)
> > >
> > > I am in favour of 3) because it seems to be the easiest way to get
> > > a literate programming system that works both ways.
> >
> > OK, that is a variant of 1).
> 
> Yes, I don't see much differences between the two approaches.

Well, perhaps there are none.  :)  See below.

 
> Otherwise I've read Felix specs and, as him, I'm in favor of a simple
> approach at first, i.e. 1) or 3). I'm not sure I've understood
> everything of the clone suggestion of Felix.

Well, as the clones idea has been officially postponed in favour of
Joris' tree obeserver concept, I won't try to elaborate right now.
 

> Felix, I do plan to work on this project, albeit not full time (I'm
> working on another free software project). I'm ready to do some
> development and tests.

Ditto.  :)

> I don't know much of C++ and Scheme. However I know about OO concepts
> (but I'm bad at OO design) and I'm using OCaml (ML-like language) on a
> regular basis.
> 
> Felix, how do you plan to do the development of this feature? Should
> we start by writting a short specification? Any idea of API/code to
> look at as a starting point?

1) Indeed, I think we should start by writing a short spec. Most
importantly to make sure that both of us have exactly the same thing
in mind when writing code. :)

2) I would propose to implement the whole thing in Scheme, for inclusion
in Scheme.

3) We could either create an export filter/serialization method, or
"override the standard load and save routines" as Joris suggested.
I think both methods differ only slightly, but I will have to dig
into the code before I see more clearly here.

4) I think using DRDs will also be of use. Perhaps we could use DRDs
to specify which nodes have verbatim content (i.e. content which should
be serialized verbatim).

5) I don't know when I have time to invest into this project. Perhaps
I will write up a quick draft of the spec monday evening for you to
change/comment on/extend. But I won't promise anything.


Regards,
Felix