On 15 Jun 2002, nix_at_esperi.demon.co.uk mused:
> However, that's not the problem here. line 295 in glibc-2.2.5's stdio.h
> reads:
> 
> /* Maximum chars of output to write in MAXLEN.  */
> extern int snprintf (char *__restrict __s, size_t __maxlen,
> 		     __const char *__restrict __format, ...)
>      __THROW __attribute__ ((__format__ (__printf__, 3, 4)));
> 
> where __THROW expands to throw() in C++ and nothing in C.
> 
> The fix is to take out -fno-exceptions, I think: g++ won't let
> exception-specifications past under -fno-exceptions.

Yes it will. I'm talking utter nonsense.

(I blame hay fever drug daze.)

I'm wondering if your compiler doesn't recognise `restrict', but that's
unlikely; even egcs recognises it. I expect some unexpected macro
expansions have chewed this code into something quite unexpected: what
does the preprocessor output for this bit of stdio.h in the context of
that file look like?

-- 
`What happened?'
                 `Nick shipped buggy code!'
                                             `Oh, no dinner for him...'