Richard Frith-Macdonald schrieb: > I'm with Fred on this one ... certainly on partially implemented > classes, but also (though less strongly) on completely empty ones. > I think there is absolutely zero risk of someone wasting loads of time > porting only to find something critical missing... as long as our > documentation does not tell lies (and little chance of it even then). > We do need to make sure that the documentation is up to date, so it > says which methods of which classes are unimplemented. > > IMO partially implemented classes tell people that there is some hope > of the classes being done in future ... or at least that the GNUstep > project would look favourably upon people contributing in those areas. > In fact it would probably be good if unimplemented methods actually > generated an NSLog explicitly asking for an implementation to be > contributed. Maybe I should add a macro to NSDebug.h to do that? > > Having a completely unimplemented class there gives us a good > placeholder for the documentation that tells people that the class is > unimplemented, and maybe what the current plans are for it. I can see > the argument here for removing the class (people aren't likely to think > the class exists if there is no trace of it), but I think that a header > file that's clearly a shell, and documentation that states that the > class is unimplemented, is equally clear. We could document such empty > classes with a note to say that someone (or nobody) is working on them, > and a pointer to the task list on the website for current status.
FWIW, I agree. Cheers, David _______________________________________________ Gnustep-dev mailing list Gnustep-dev@gnu.org http://lists.gnu.org/mailman/listinfo/gnustep-dev
