Re: giomm API review needed

[Murray Cumming <murrayc murrayc com>]

On Sat, 2008-02-23 at 14:28 +0100, Marko Anastasov wrote:
> On Sat, Feb 23, 2008 at 1:56 PM, Murray Cumming <murrayc murrayc com> wrote:
> >
> >  On Wed, 2008-02-20 at 23:04 +0100, Marko Anastasov wrote:
> >  > - optionally include unix headers
> >
> >  Do you mean from giomm.h?
> 
> Yes, although now after looking at gtkmm.h it seems that they
> should just be included unconditionally?

Yes, gtkmm.h includes, for instance, pagesetupdialogunix.h. But that
doesn't seem wise. I don't think that header would even exist on
Windows. I guess someone will need to fix that sometime.
 
Let's require people to include the unix* headers explicitly, without
including glibmm.h, for now. We can do something more clever in future
if we like.

> >  Does anyone know of anything else?
> 
> I suppose to decide about the additional InputStream::read() overload.
> Since we already have OutputStream::write(std::string, ...), I would now
> say that for consistency there should be a read(std::string, gsize count).
> Although I personally do not find them alone helpful enough.

Please make sure there is a TODO about that in the .hg file.

I think it will have to wait for glibmm 2.17/18.

-- 
murrayc murrayc com
www.murrayc.com
www.openismus.com
>