dolfin team mailing list archive

[Johan Jansson <johanjan@xxxxxxxxxxxxxxxx>]

On Wed, Aug 09, 2006 at 01:33:59PM +0200, Anders Logg wrote:
> On Wed, Aug 09, 2006 at 01:06:08PM +0200, Johan Jansson wrote:
> > On Wed, Aug 09, 2006 at 12:26:03PM +0200, Anders Logg wrote:
> > > Is there a way to recursively include stuff in SWIG so we don't need
> > > to list every single file in dolfin.i? It should be enough to include
> > > dolfin_la.h for example.
> > > 
> > > /Anders
> > 
> > Yes, there's an option "-includeall" which follows all include
> > statements (not sure if it recurses beyond that). The problem is that
> > then SWIG will start parsing and wrapping almost everything,
> > i.e. system headers, PETSc, uBlas, etc. This will lead to an error
> > (SWIG can't parse the uBlas headers for example), or at least to a
> > very bloated PyDOLFIN.
> > 
> > I think the best solution right now is to specify the headers
> > individually.
> > 
> >   Johan
> 
> Is there no way to just do it a one-level recursion, so we can include
> the dolfin_foo.h files?
> 
> It would be simpler to maintain if we don't need to edit both the
> dolfin_foo.h files and dolfin.i.
> 
> /Anders

I'll check if it's a one-level recursion or not. The manual only says
that unless -includeall is specified, the #include statement is
ignored.

  Johan