instant team mailing list archive

["Martin Sandve Alnæs" <martinal@xxxxxxxxx>]

On Tue, Jan 13, 2009 at 11:11 AM, Johan Hake <hake@xxxxxxxxx> wrote:
> On Tuesday 13 January 2009 10:56:16 Martin Sandve Alnæs wrote:
>> On Tue, Jan 13, 2009 at 10:51 AM, Kent Andre <kent-and@xxxxxxxxx> wrote:
>> > On ti., 2009-01-13 at 10:14 +0100, Johan Hake wrote:
>> >> On Tuesday 13 January 2009 09:59:30 Kent Andre wrote:
>> >> > As always, improved documentation and better tests.
>> >> > Anything else ?
>> >>
>> >> I have expanded the kwargs in build() with swig_include_dirs, wo be able
>> >> to %import other swig modules. I needed this for the shared_ptr stuff.
>> >> This will not break anything, so I can send a patch now or I can wait.
>> >> You deside.
>> >>
>> >> Johan
>> >
>> > I've used swigopt to set swig_include_dirs, but it may not be obvious.
>> >
>> > Kent
>>
>> I think that's fine, and we probably don't want to duplicate all
>> possible commandline options...
>
> The problem with swigargs (not swigopt) is that this comes with a lot
> predfined options, which all make sense, e.g., -c++, -I. (well, we could
> skip -fcompact and -small, makeing the produced wrapper code readable).
>
> When I then put e.g., [-I/path/to/something] as a swigargs it will exclude the
> rest. This is not a big issue, but it was not obvious that I should put
> include directories in swig_args.
>
> Having one argument for general swig options and one for include directories
> didn't seem too overkill for me.
>
> Johan

Ok by me. I guess this is consistent with the C++ includes vs flags as well.

Martin