[Bug target/63651] Lot of failures in obj(c|-c++) with yosemite

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63651

--- Comment #17 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to mrs@gcc.gnu.org from comment #16)
> We can fixincludes NS_BLOCKS_AVAILABLE support back in, this would disappear
> interfaces that cannot be supported.  In the past, this was a safe thing to
> do, and might well be still safe wrt the runtime.
> 
> Deeper language issues would likely need someone to do real work.  No really
> nice fix for that other than someone who wanted to do the work stepping
> forward.  Until then, SDK support for older OSes might be the old way to get
> code compiled on newer systems.
> 
> We should be able to steal code from MIT style runtimes to put into newer
> systems, if we can get FSF approval for incorporating code they don't own. 
> This should be easy enough, we don't vend sell or ship a competing abi
> compatible runtime, so, bundling one I think should be trivial, if we want
> to.

two things here:
(1) making FSF ObjC useful again on Darwin - the only solution there
realistically is to implement blocks [TBH, this applies outside of ObjC too]. 
Likely there are two candidates for that job - mrs and ids .. ;) .. if there
were some way to split the task up..  I believe this is rapidly becoming a
show-stopper for FSF GCC on darwin. :(

(2) Having a non-fragile ObjC library on Linux (and/or other interested *nix
hosts).  My understanding is that David Chisnall's libobjc2 might fit the bill
for this (BSD license) ..