On Fri, 2003-06-06 at 11:57, Jason Merrill wrote: > Currently, passing something which is not an lvalue will cause a > gimplification failure. We ought to give a better diagnostic sooner; > perhaps rth's 3.3.1 patch will do that. > Ah, that would be the new failure in gcc.dg/20010516-1.c, right? Diego.