This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RFA][objc] Identifier cleanup
Ziemowit Laski wrote:
David,
This is a good idea, but perhaps I can convince you wait about a week,
and then do this on the objc-improvements-branch instead? :-)
Zem,
Well, you've got the OK to merge the branch (even in stage 2 IIRC?).
After this is in, (either tree,) I'll still need about a week for the
other (2?) patches to be cleanly split, bootstrapped and tested. I'm
just a bit worried about the stage 2 deadline. But if you say I have
that time, then I'm fine with it.
I just hope that the merge will be smooth and this doesn't have to be
backed out. We're still experiencing minor fallout from the last one
due to lacking test cases. (Silent change in the encoding of selector
types. But we haven't figured out whether the change is good or bad
yet, so there is no bug report yet. What was bad is that it was changed
silently and executables compiled with different compiler versions now
have conflicting type info. Also the ChangeLogs were of no help in
hunting that issue down.)
Among
other things, my current objc-improvements-branch mods (not all of them
yet committed, due to some DejaGNU regressions) create builtin typedefs
for 'id' and 'Class', so that neither needs to be a keyword. I've also
folded get_object_reference and get_static_reference into a single routine.
AFAICT Class was never a keyword but I would have either had to make one
or remove id also. I also folded those two functions so yes this
sounds very much in line with my changes.
Cheers,
David