This is the mail archive of the
mailing list for the GCC project.
Re: Patch: fix ObjC regression (sort of)
On Dec 21, 2004, at 3:19 PM, Ziemowit Laski wrote:
If it is not a regression, I think it should wait for gcc 4.1. I take
it this is
already in apple-ppc-branch, right?
Depends on how you define regression. Yes, it's in apple-ppc-branch.
On 21 Dec 2004, at 14.24, Dale Johannesen wrote:
The ObjC representation of calls to objc_msgSend was
changed to use OBJ_TYPE_REF rather than NOP_EXPR. Nobody
taught get_callee_fndecl about this change, which means that such
calls are treated as calls through a pointer. This is wrong and a
However, I don't know of any user-visible effect this has, so I can't
provide a testcase, and you may consider this isn't suitable for
(I know about it because it broke a local Apple patch.) Built on
darwin and ObjC testsuite run (change is local to ObjC bits).
2004-12-21 Dale Johannesen <email@example.com>
* objc/objc-act.c (objc_get_callee_fndecl): New.
* objc/objc-act.h: Declare it.
* objc/objc-lang.c (LANG_HOOKS_GET_CALLEE_FNDECL): Use it.