g-wrap to produce alternative output?

[Linas Vepstas]

On Fri, Aug 03, 2001 at 04:55:10PM -0700, Dave Peticolas was heard to remark:
> On 29 Jul 2001 19:01:00 +0000, Tripp Lilley wrote:
> > Howdy...
> > 
> > I want to write Python extensions to GnuCash. Before g-wrap, I suppose I
> > would have added Python output to the SWIG wrappers. After g-wrap, well :)
> > 
> > Anyway, I started poking around with g-wrap a little bit, and it seems
> > like it ought to be a tractable problem to generate Python extensions from
> > what g-wrap is already given (ie: gnc.gwp), rather than either re-wrap
> > gnc directly in Python, or pick the SWIG remains back up.
> > 
> > The fourth alternative, I suppose, would be to work towards a "canonical"
> > representation of the API that could then be coerced into producing inputs
> > for g-wrap, SWIG, or what have you. However, this seems an awful lot like
> > reinventing SWIG (which, I understand, you dropped because of its poor
> > Scheme output).
> 
> Swig was dropped because our use of it was broken due to changes in 
> the engine API, and nobody was stepping up to keep it up to date.
> However, Linas has recently resurrected it and updated it in the
> development tree.

Basically, I got the perl bindings going.  One thing I did *not* do
was to create typemaps between GLists and perl lists.  Presumably,
one would want to do this so that one could get at e.g. the list
of accounts or transactions that some function is returning ...

Presumably, you'd need to do the same for python ...

--linas


-- 
I'm very PUBLIC-MINDED, I'm helping a NIGERIAN get his $25,000,000 back!