kicad-developers team mailing list archive

[Dick Hollenbeck <dick@xxxxxxxxxxx>]

On 08/09/2012 09:07 AM, Dick Hollenbeck wrote:
> http://www.swig.org/Doc1.3/Python.html#Python_nn17
>
> Suggests that named objects are created in python by swig for each unique enum value.
>
> Don't know that there is any problem if *.i is done right.


I think this means including more C++ headers into the *.i files to pick up the C++ enums
and defines.

As a means of surgically controlling which specific items are exported by SWIG, I think
this means we have to tolerate at least the


#ifndef SWIG
#endif

construct.



SWIG should not be #define-d when C++ is compiling, so one way to tell the swig compiler
to ignore something is the

#ifndef SWIG
#endif


construct, like you see at line 80 in dsnlexer.h, which I put in there back when I was young.

This much, seems like a reasonable concession to use SWIG, but any more of an imposition
we need to talk about, and I would want to get Jean-Pierre's opinion and Waynes in that,
should more be needed.

The tail should not start wagging the dog.