Re: GNUstep roadmap (was Re: [Suggestion] GNUstep-test for quality control)

[Pete French]

> Actually, I don't know exactly, what GSArray is specifically for. As 
> far as I see, it is a real class. CFArray therefore is a non OO 
> implementation of an array and it serves as the base for a real 
> NSArray. NSArray is mostly a wrapper for a CFArray.

Is this on OSX ? I thought that might be how its done - or else
make CFArray a wrapper for NSArray (which is how I considered doing
it on GNUstep)

> Not everything is interchangeable. If I recall that right, even the 
> majority is not.

I think its just the ones which match FoundationKit objects, I havent
used the bridging myself other than for simple things.

> It could be. Depending, on how this kind soul implements it.

:-) I has assumed it was a straight port of the CF source code. An idea
would be to simply port those bits of the CF source which dont bridge, and then
make those which do into a wrapper round the existing GNUstep implementations
of NSArray etc...

So, how *are* you implementing it ?

> I do agree with that. But I would like to repeat here, that when we got 
> it into the minds of those programmers, that GS is a viable possibility 
> to open the Windows and Linux market afterwards, that they would try 
> not to use CF. The few moments, where Cocoa doesn't offer an 

You are more optimistic than me in that case! I also think its a catch 22
situation - that the more special cases the OSX people need to take on board in
order to use GNUstep, the less likely they are to do so. Conversely if
they think theres no difference then they are more likely to take it into
account as they think its zero work. Developers are very lazy ppl :-)

Just browsing through Apples CF code at the moment actually - interesting
that there are the occasional ifdefs for Linux and FreeBSD already (at
leats in one of the header files) so obviously they have an interest in
porting this themselevs...

-bat.