Re: RFD : libg++/gcc/egcs upgrades needed for libc6 (READ ME)
On Sun, Feb 22, 1998 at 04:27:53PM +0000, James Troup wrote:
> Galen Hazelwood <galenh@micron.net> writes:
> Well not much I can do about if I do, is there? :-P
>
> > > I'm not trying to be argumentative, I just genuinely don't understand
> > > this urgent need to switch compilers.
> >
> > I'm pretty sure this was brought up in debian-devel. Check the
> > archives for subjects matching the one on this message.
>
> Well, while not as reliable as Guy's archives, my memory archives
> can't remember anyone presenting valid reasons as to why exactly we
> need egcs so badly as a C compiler and we surely must base the
> compiler choice on that? C++, g77 & non-i386 architectures[1] are
> important, but not as important as the C compiler IMO. (Compare and
> contrast number of C vs. C++|g77 programs in archive then compare and
> contrast number of packages-working-with-2.7.2.3
> vs. packages-requiring-egcs)
>
> > Are you seriously telling me that the m68k leaders are publicly
> > repudiating egcs?
>
> As used for compiling a `stable' distribution; yes, and FWIW I agree
> with him.
>
> "egcs is an experimental step in the development of GCC, the GNU C
> compiler."[2]
>
> > I personally think any distribution compiling things with 2.8 is
> > more broken, [...]
>
> Jes wasn't suggesting 2.8; rather 2.7.x.[3]
>
> > Events have given me a chance to rethink my decision. I'm currently
> > waiting for gcc 2.8.1 and egcs 1.0.2 to appear. If 2.8.1 will
> > compile the kernel (through behavior or documentation changes ;), I
> > will consider it back in the running. I won't ship a compiler with
> > debian which can't compile the kernel.
>
> I agree, it would be very broken to do so. So why not stick with a
> compiler we *know* compiles the kernel (2.7.2.3)? :-)
Well, it may compile the kernel, but it is broken in respect to C++.
Although C++ may not be so important as C at the moment, the exception
handling (for example) is unusable in gcc 2.7.x. It is much better in egcs
and maybe in gcc 2.8 (which I did not try).
I did not follow the thread, but if you talk about which compiler to include
in hamm, then I would suggest the next useful combination of egcs and gcc >=2.8
> > I like the egcs development model, and think that using it as our
> > default compiler would give them a lot of moral support.
>
> IMHO we shouldn't sacrifice the safety of our C compiler for the
> dubious goal of giving the egcs team moral support.
The moral support is a side effect. The goal is to put pressure on the gcc
group to release stable versions earlier, as I see it.
Marcus
> > Speak soon or forever hold your peace...
>
> I really didn't intend to make this much of an issue out of it, which
> is why this was {originally} private mail. If you want to make egcs
> the default Debian compiler, that won't actually affect m68k, since
> we'll just ``be slow to compile egcs'' and carry on using gcc.
>
> [1] Non-i386 architectures can and will use their own compiler
> (e.g. everyone but m68k already is IIRC).
> [2] <URL:http://www.cygnus.com/egcs/>
> "egcs project home page"
> [3] Though for m68k, 2.8 probably makes more sense, it doesn't have
> the same problems as i386 with regards to kernel compilation, and
> has more (m68k) development weight behind it than egcs.
>
> --
> James
>
>
> --
> TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
> debian-devel-request@lists.debian.org .
> Trouble? e-mail to templin@bucknell.edu .
>
--
"Rhubarb is no Egyptian god." Debian GNU/Linux finger brinkmd@
Marcus Brinkmann http://www.debian.org master.debian.org
Marcus.Brinkmann@ruhr-uni-bochum.de for public PGP Key
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/ PGP Key ID 36E7CD09
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org .
Trouble? e-mail to templin@bucknell.edu .
Reply to: