weird SWIG syntax error with "inline" keyword

[Mark Jenkins]

I wrote:
> I also worry that the -Dinline approach could come back to haunt us if
> the word inline was to appear in a header where it wasn't simply an
> ignorable C keyword, eg; int inline_skating_shoes();, though I'll
> concede this seems unlikely.

A baseless fear, as the pre-processor doesn't behave this way, this
would only act on the inline keyword, or functions named "inline".

On second thought, adding -Dinline doesn't sound like a bad idea, as its
only downside is extra code, but on the upside it means less is required
for other developers to get involved. I think that makes it worthwhile.

Christian wrote:
> We either need to upgrade this
> (on trunk) to 1.3.31 or we need to add a workaround for versions lower
> than 1.3.31, which would mean adding "-Dinline=" to SWIG_PYTHON_OPT.

If we're going to do this, we shouldn't just add it to SWIG_PYTHON_OPT,
but somewhere that will be used by all invocations of swig within
GnuCash, as this isn't a swig -python specific problem. Someone else may
wish to run gnc-numeric.h through swig sometime. (when they attempt to
write Java bindings, *shudder*. )


Mark