Pending GCC-3.1 upgrade - do or don't

Ian Molton spyro at armlinux.org
Tue May 7 14:43:16 PDT 2002


On Tue, 7 May 2002 15:21:22 -0400
Gerard Beekmans <gerard at linuxfromscratch.org> wrote:

>I need proof. Show me a package that
> doesn't compile properly when compiled with gcc-3.1 (whatever is in
> CVS now, not gcc-3.0.4) and convince me that it's gcc's fault, not
> attributed to bad coding skills by that package's programmer.

Thats a straw-man argument, unless you are proposing LFS will use CVS
versions of GCC. (oh god, please no!)

> I'm looking at the large picture:
> the thousands of people who would potentially upgrade.

Why the rush to have a bigger version number?!

> I cannot add to the book "This gcc version may not work for you. We
> don't know when it doesn't, just build an entire LFS system and if it
> doesn't work, start over again with an older GCC version and try
> again". I hope you can see that simply isn't good enough.

no. of course not.

Better would be "LFS recommends the use of stable, solid, tested,
compiler version X.
LFS has also been successfully built using experimental compiler Y, and
some people have found it to be stable, but this is a relatively
untrodden path, and not recommended unless you are skilled in debugging
the subtle build problems that may arise.
 
> People keep saying there are all kinds of things wrong with GCC-3.1.
> Thus far I only know that _something_ may be wrong, I still don't know
> _what_ is wrong.

Would you use a parachute under those circumstances?

> How is anybody supposed to know if gcc-3.1 is
> going to be so terrible? It's not even released yet so nobody can
> actually know. Whatever is in CVS may change between now and the
> release.

Try subscribing to the GCC mailinglist. Even from my light reading of
it, I see numerous bugs being discussed, many with people saying 'oh god
thats bad, and even worse in the current branch...'.

> Examples guys, I want to see examples of bad gcc-3 behaviour

why? do you not believe its authors? Would you trust the latest product
of a parachute maker if it was labelled 'might not open' ?
-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message



More information about the lfs-dev mailing list