Hi, This will be my last mail in this exchange, since it looks like it's a dead end. > I don't know if we've broken it somehow, or (more likely) a bug > in the HP compiler. Well, then, it's not only in the HP compiler, but in gcc-3.4.4 and gcc-4.0.2 too. Which leads me to think it's a general ia64(-hpux) problem in the sources. FX