On 12/02/12 06:41, Ian Lance Taylor wrote: >> This syntax works fine in gcc 3.4 but seems to have disappeared in gcc4, >> where it complains that the symbol is aliased to undefined symbol. > I don't see any evidence that this actually worked in gcc 3.4. In 3.4 > it would make Localfoo an alias for a symbol named literally > "Foo.Foodata". it would not make Localfoo an alias for the Foodata > field of the global variable Foo. I think gcc4 is kindly giving you an > error for something that never actually worked. > > Ian That's odd. Even if the compiler didn't check it, I would have expected a linker error in such case. Although I only tried to compile it, not to actually use it, maybe some optimization was removing all the symbols altogether, making it look as if it worked. I'll check tomorrow and report back. Thanks for looking at it, Ian.
