Subj : Re: Non-strictly-conforming and unspecified versus undefined behavior
To   : comp.programming.threads,comp.std.c
From : Alexander Terekhov
Date : Sat Feb 19 2005 11:49 pm


David Schwartz wrote:
[...]
> is logically impossible for a standard to impose requirements on programs
> that invoke undefined behavior.

The silly thing you seem to like so much can be accomplished with 
rvalues, not lvalues as __thread (thread storage duration) extension 
does.

regards,
alexander.

.