------- Additional Comments From davids at webmaster dot com 2005-02-24 20:42 ------- >The only thing which would change when you remove the automatically >inserted locking is that some programs which used to work are now >broken, and that some other programs which used to deadlock now >invoke undefined behavior from entering an initializer recursively.
First, if we're talking about pthreads programs, which is the only case I'm suggesting removing the locking for, then those programs are already broken. The pthreads standard requires locks when data may be changed in one thread and accessed in another. Entering an initializer recursively has always been undefined behavior. Any code that might do that is broken. In any event, the entire thrust of this argument is bogus. If GCC/G++ are going to have non-portable features that make code work when they're enabled and break when they're disabled, they definitely should not be on by default. That they hide bugs in code that claims standards compliance on one platform, but allow them to fail on another, is an argument *against* the option, not for it. (Or are you seriously arguing that the C++ standard and the POSIX standard *require* this behavior?) -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20099
