> Some of these are Alpha specific problems,
Oh, no doubt about it - my alpha source tree is full of small hacks
because I'm too lazy to get a real egcs (egcs on my alpha is really
broken).
This reminds me, __builtin_return_address is real pain in places, is
there anyone where absolutely need this, it seems evil at best and it
might be nice to eradicate it completely from the kernel.
> in other cases it is often bugs in the c++ code itself which g++
> used to silently allow, but now with how closely egcs follows the
> ANSI c++ standard (much closer than any commercial offering I know
> of) you will often witness c++ code which needs to be cleaned up.
I can't comment on this other than to say there are reports of large
libraries being ported to linux/axp which build and work on other
platforms (eg. Solaris and IRIX) and do not under linux/AXP.
I don't think these are publicly available projects so its not going
to be that easy to track this down using second hand information, the
actual people experiencing the problems are going to have to be
cooperative.
> And btw, broken egcs c++ issues aren't going to make as much
> headway to get fixed if discussed only on redhat-axp, I'd take it
> also to one of the egcs lists where the maintainers are more likely
> to be listening :-)
Oh, I absolutely agree, but I'm not 100% sure all the problems that
have been reported are in fact egcs problems, some on them are
certainly binutils problems/limitations.
No doubt rth has also read these comments and hopefully will pass on
any relevant parts to the right people, he's also going to be much
better qualified than myself and spotting what's an egcs problem and
what's not.
It also appears that many alpha people are using egcs where as this
may not be the case on other architectures (although sparc64 uses
egcs too, no?)
-Chris
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/