Re: Python transformer for Guile?

[sen_ml at eccosys dot com]

earlier i wrote:
>    one thing i have wondered about in the past is whether the parsing
>    code used to implement the source language (in this case python) can't
>    be re-used for the task -- at least from a technical perspective.

neil> I have wondered exactly this for Elisp.  It seems like it should work:
neil> all you have to do (using the Elisp case as an example - I don't know
neil> Python) is branch the reader code at the point where it creates a
neil> LispObject for whatever it has just read so that it creates either a
neil> LispObject or a SCM.  I suspect that the main problem would be code
neil> management - how do you make sure that the two branches stay up to
neil> date with each other?  (This problem applies independently of whether
neil> the Guile code is in the Emacs codebase or in a copy of Emacs' read.c
neil> in the Guile codebase.)

neil> Or am I being seriously naive here?

well, i don't think i have a good answer for you because i was
wondering the same thing [whether i was being naive, not you of
course!] :-)

as far as the code management goes, it might be nice if both the
source language and the translator used a common library for the
parsing code.

if i understood one of lalo martins' comments correctly, it isn't
necessarily going to be easy to re-use the parsing code for an
arbitrary implementation of a source language.

i don't suppose language implementors can be encourage to implement
their languages so that the parsing portions can be reused nicely by
translators...