On 11/21/2011 02:47 AM, Andrew Haley wrote: > On 11/21/2011 04:13 AM, Jeff Law wrote: >> This is (arguably) a glibc issue. I'm still investigating. >> >> Attached you'll find the hack from the gcc46 srpms we're using to work >> around the problem right now. > > Why is isspace() marked throw, anyway?
Not throw as in throwing, but throw() as in no-throwing. IMO it's glibc confusion about what C++ throw() means as opposed to __attribute__((__nothrow__)). The later is merely an assertion by the programmer; the former is a request for the compiler to ensure that no exceptions actually propagate at runtime. r~
