|
From: Julian S. <js...@ac...> - 2008-12-19 09:27:26
|
> That seems to work for me anyway... Yeh, me too (playing around on an ancient RedHat 9 VM). But doesn't this rather rely on the assumption that gcc will fold the underlying expression, &((type*)0)->field, to a constant at compile time? I guess gcc wouldn't be able to satisfy the "n" (this arg is a constant) constraint in the asm if it didn't. Oh well. J |