[Bug inline-asm/11564] [3.3/3.4 regression] no longer compilable inline asm breaks linux kernel compile

["rth at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11564


rth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |INVALID


------- Additional Comments From rth at gcc dot gnu dot org  2003-07-17 21:56 -------
<rth> the problem with this (and the whole notion that ? can be an lvalue) is
that the two arms of the ?: do not have the same type.
<franzo> but that is exactly the feature that prevents a lot of warnings if we
have a long long temporary in the size 8 get_user case
<rth> frankly, the manual does not support the usage as given.  it mentions only
(a?b:c)=5 --> (a?b=5:c=5).
<rth> i'm not tempted to extend the semantics of this already questionable feature.
<franzo> hmm, ok. then we can change evdev.c to use a temporary and hope not too
many other places in the kernel do that