Re: GNUstep Base OpenStep Compliance

[David Ayers]

Richard Frith-Macdonald wrote:

> [snip]  and updated your OpenStep compliance document to say that the 
> only non-conforming bits are -
>
> > NSBTreeBlock
> > NSBTreeCursor
> > NSByteStore
> > NSByteStoreFile
This is Great!

And all together it *is* a tremendous achievement!!  GNUstep effectively 
has OpenStep compliance for base!  And not only does it comply to the 
written specification, it also does this in a meaningful, (mostly?) 
consistent way.  So that part of the "Mission Statement" is achieved!

Yet even if we could say the same for gui, I don't think, this is what 
really prompted the discussion.  In my view the discussion pointed at:

(some people would have ordered this differently ;-) ... )
- OPENSTEP 4.2 compatibility*
- Cocoa current compatibility*
- stability of "implied behavior"
*(API incl. implied behavior as needed to port code from OPENSTEP and as 
needed to implement code which can be cross-compiled on Cocoa.)

All three go beyond specification compliance.  They may even be 
complementary goals.  Especially the last one (I believe the one that 
has been causing the frustration of some developers targeting GNUstep) 
is effectively impossible to achieve, if we continue to achieve the 
first two.

I would like the "Mission Statement" to include at least the first two.  
Plus I would like guidelines, and or mechanisms in place to handle 
conflicts between the first two.  I believe that even though the last 
one might have prompted the discussion, I agree with Nicola that 
"stability" is implied and shouldn't be part of an official statement, 
realizing that as long as we agree to achieving the first to but 
haven't, the latter will probably just not happen.

If we can agree on this addition to the "Mission Statement", then I 
would actually consider obtaining an OPENSTEP 4.2 for Mach and OS X. 
(God forbid, I meant not to buy any proprietary software anymore :-) )  
And start writing more regression test that run on OPENSTEP 4.2 
(Enterprise/Mach), OS X and GNUstep.

Putting the "Mission Statement" aside, those of us who are frustrated by 
the latter point, should become more concrete about which changes are 
causing the frustrations and how they should be handled in the future.  
As long as we're just complaining, we can't expect to see constructive 
change.

My suggestion would be to start posting ChangeLog entries (possibly 
including patches), so that people have it easier to follow what's going 
on.  I've been doing that, but as no one joined in, I guess, I've been 
annoying people more than anything else.  Maybe we should have a 
seperate patches-gnustep / commit-gnustep or whatever mailing list, 
possibly separate ones for core and other projects dedicated to this.  
Anyone have better suggestions?

Cheers,
Dave