Re: [Swig-user] Rewrite SWIG to Python? (Was: Announce - SWIG-1.3.30)

[Mike R. <mik...@no...>]

>>>>> " " == Henning Thielemann <le...@he...> writes:

[snip]

     > I see, I did a big mistake: If there would be consensus on
     > which language to use, then there wouldn't be a wide range of
     > scripting languages, and there wouldn't be any SWIG. So
     > proposing a language to which to rewrite SWIG is
     > contradictory. Even more, starting a discussion about rewriting
     > SWIG in a different language can only lead to a stalemate.

  Many years ago, there was a suggestion on the SWIG mailing list that
SWIG should be extend-able by itself.  In other words, why cant SWIG be
extended by using the scripting languages it creates wrappers for?
This way, parts of SWIG could be implemented in many different
languages.  

  I think this is a cool idea that should be revisited.  It sorta
makes sense to me that some parts of SWIG that create wrappers for
python might be implemented in python.  And perl in perl, Tcl in Tcl,
Java in Java, etc.  

  I'm not a SWIG developer, but I have looked inside
SWIG from time to time in order to debug things.  To, me it always
looks a bit clunky sorting out how the C++ generates python.  One has
to hunt amongst the C++ to find out how it is generating the python
wrapper code.  I think it would be nice if this type of stuff was in
python.   And of course, the same would apply to other languages.   I
would still expect that some core part of SWIG would remain in a
single language (just to prevent chaos).

Mike Romberg