Markus Baeurle wrote:
>
> Doesn't seem so.
> A hello-world-type program still gives "RRS text relocation" (Btw,
> what does this mean?) for cout, but results in a working binary.

That's not a problem. There's absolutely no reason for those messages to
be printed out, though, I'd have thought they were removed by now...

> // Kleines Programm mit setw() und rechtsbuendiger Ausgabe
> 
> #include <iostream.h>
> #include <iomanip.h>
> 
> int main (void)
> {
>         int i = 8;
> 
>         cout << setiosflags (ios::left) << setw(22) << "Zahl:" << i << endl;
>         return (0);
> }
> 
> which gives this linker error message:
> 
>  ld -e start -dc -dp /usr/lib/crt0.o -L/usr/libexec /var/tmp/cc00231a1.o -lg++ -lstdc++ -lm -lgcc -lc -lgcc
> /var/tmp/cc00231a1.o: Undefined symbol `_setw__Fi' referenced from text
> segment
> /var/tmp/cc00231a1.o: Undefined symbol `_setiosflags__FUl' referenced
> from text segment
> 
> I applied the two older patches some time ago and then the latest patch
> you mentioned above.

Hmm if I compile with gcc I get this but if I compile with g++ it works
fine...

> I'd really appreciate if it was checked in. I don't get working kernels
> with the most recent csc.o (posted to this mailinglist in December
> IIRC).

Mark can you upload a copy of the driver somewhere? It wont be checked
in till Scott's done the rework, I don't think. Scott?

> But this law is still true: "Software is a gas. It will always take up
> all available space." ;-)

:)

	N.