instant team mailing list archive

[Johan Hake <hake@xxxxxxxxx>]

On Tuesday 13 January 2009 10:39:48 Martin Sandve Alnæs wrote:
> Just thinking loudly here:
>
>
> This is a declaration:
>
>   void foo(int bar);
>
> This is a definition:
>
>   void foo(int bar)
>   {
>     cout << bar;
>   }
>
>
> SWIG only needs the declaration, but can take the definition as well
> (just throws away the garbage).
>
> Since additional_definitions is included together with
> additional_declarations in the "non-%{}" section, additional_declarations
> is intended for
> stuff that is only
> declared for wrapping by SWIG while additional_definitions is intended for
> stuff that needs to be compiled into the SWIG module _and_ wrapped by SWIG.
>
> additional_declarations can depend on additional_definitions, but not
> the other way around.
>
> Martin
>
>
>
> On Tue, Jan 13, 2009 at 10:33 AM, Martin Sandve Alnæs
>
> <martinal@xxxxxxxxx> wrote:
> > Definitions are also included in the %{ %} section.
> > Definitions and declarations have specific meanings in C/C++,
> > but I'm not sure how clear that distinction is in their use here:
> >
> >    interface_string = reindent("""
> >        %%module  %(modulename)s
> >        //%%module (directors="1") %(modulename)s
> >
> >        //%%feature("director");
> >
> >        %%{
> >        #include <iostream>
> >        %(additional_definitions)s
> >        %(system_headers_code)s
> >        %(local_headers_code)s
> >        %(wrap_headers_code1)s
> >        %(code)s
> >        %%}
> >
> >        //%%feature("autodoc", "1");
> >        %%init%%{
> >        %(init_code)s
> >        %%}
> >
> >        %(additional_definitions)s
> >        %(additional_declarations)s
> >        %(wrap_headers_code2)s
> >        //%(typemaps)s
> >        %(code)s;
> >
> >        """ % locals())
> >
> >
> > Martin

Mak sense.

Johan

> >
> > On Tue, Jan 13, 2009 at 10:17 AM, Johan Hake <hake@xxxxxxxxx> wrote:
> >> On Tuesday 13 January 2009 10:10:26 Martin Sandve Alnæs wrote:
> >>> There is some small stuff in the TODO file.
> >>> And there are some TODO and FIXMEs in the code.
> >>> This is probably overlapping.
> >>> In particular, there are some old arguments
> >>> that may be removed or must be explained.
> >>
> >> Also, what is the difference between additional_declarations and
> >> additional_definitions? It seems that both gets included before the
> >> main %include of the the source, in the constructoed interface file.
> >> Whould it be instructive to have a pre_declaration and a a
> >> post_declaration as we have both in pycc and dolfin?
> >>
> >> Johan
> >>
> >>> TODO:
> >>> - Fix arguments not used in setup.py
> >>>
> >>> - Remove unused csrcs stuff.
> >>>
> >>> - Explain arrays argument.
> >>>
> >>> - Use object files argument.
> >>>
> >>> - Improve documentation
> >>>
> >>> - Add tests for all variants of cache mechanisms available now
> >>>
> >>> - Clean up imports, don't use import *
> >>>
> >>>
> >>> martinal@martinal-xps:~/dev/fenics/instant/src/instant$ egrep FIXME *
> >>> build.py:          FIXME: Describe this correctly. Tests pass arrays
> >>> of arrays of strings.
> >>> build.py:        instant_error("FIXME: setup.py doesn't use the C
> >>> sources.") codegeneration.py:      - additional_definitions (FIXME:
> >>> comment)
> >>> codegeneration.py:      - additional_declarations (FIXME: comment)
> >>> codegeneration.py:      - arrays (FIXME: comment)
> >>> codegeneration.py:    #instant_warning("FIXME: Not using csrcs in
> >>> write_setupfile().")
> >>> codegeneration.py:    arrays = [] # FIXME: Example input here
> >>> martinal@martinal-xps:~/dev/fenics/instant/src/instant$ egrep TODO *
> >>> build.py:# TODO: Import only the official interface
> >>> build.py:        # TODO: Error instead? Indicates race condition
> >>> build.py:        - If you want to compile the files yourself. TODO:
> >>> Not yet supported.
> >>> cache.py:# TODO: We could make this an argument, but it's used
> >>> indirectly several places so take care.
> >>> __init__.py:# TODO: Import only the official interface
> >>> inlining.py:    # TODO: Something more robust? Regexp?
> >>> inlining.py:    # TODO: This isn't very general!
> >>> inlining.py:    # TODO: This isn't very general!
> >>> inlining.py:    # TODO: Is the second and third path here necessary?
> >>> inlining.py:    # TODO: Is the second and third path here necessary?
> >>>
> >>>
> >>> Martin
> >>>
> >>> On Tue, Jan 13, 2009 at 9:59 AM, Kent Andre <kent-and@xxxxxxxxx> wrote:
> >>> > As always, improved documentation and better tests.
> >>> > Anything else ?
> >>> >
> >>> > Kent
> >>> >
> >>> > _______________________________________________
> >>> > Instant-dev mailing list
> >>> > Instant-dev@xxxxxxxxxx
> >>> > http://fenics.org/mailman/listinfo/instant-dev
> >>>
> >>> _______________________________________________
> >>> Instant-dev mailing list
> >>> Instant-dev@xxxxxxxxxx
> >>> http://fenics.org/mailman/listinfo/instant-dev