Re: Building gnuStep on panther

[David Ayers]

Fred Kiefer wrote:
> David Ayers wrote:
>
> > Well not necessarily, but yes.  In fact the problem is that, no one 
> > has gone through -gui and -back yet to add:
> >
> > #include <GNUstepBase/GSObjCRuntime.h>
> >
> > where necessary which includes further headers that will do the 
> > mapping from GNU runtime functions to Apple's runtime.  But note that 
> > there isn't always a (useful) mapping available.  When that's the case 
> > we try to provide a GS(ObjC)SomeThingFunction in GSObjCRuntime.h to 
> > abstract that difference.
> >
> > I'm not too involved with -gui but I just put this at the end of my 
> > TODO list.  I'd be very glad if someone else could look into this as I 
> > cannot say that I'll be able to deal with this any time soon.
>
>
> Hi David,
>
> could you please be a bit more explicit, where you think GUI is doing 
> things wrong? The only method I could find that uses strange OBJcRuntime 
> fucntions was [NSGraphicsContext subclassResponsibility:] and even that 
> seemed ok. I hope that all the complicated things get handled by base, 
> so gui is not concerbed by them at all.

Hello Fred,

I've just grepped through -gui and it seems you're right.  There are 
only a few places where objc_malloc/_realloc/_free are used 
(GSKeyBinding*/tiff.m).  I guess one would need to investigate further 
where these other objc_ symbols are used (or if the result from 
including the correct headers with the wrong defines being active.)

Cheers,
David