Re: gcc 4.3: when will it become standard compiler?

On Fri, Jan 9, 2009 at 2:23 AM, Svein Skogen (List Mail Account)
<svein-listmail_at_stillbilde.net> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Christoph Mallon wrote:
>> O. Hartmann schrieb:
>>> When will gcc 4.3 incorporated in FreeBSD 8 and become the standard
>>> compiler suite? We figured out that gcc 4.3 does have a speed gain in
>>> some numerical code of 3 - 8 % and I guess we can use this in the basic
>>> OS as well ...
>>
>> Number crunching has a totally different execution profile than basic
>> operating system services. Gains in one area cannot simply be
>> transferred to the other.
>
> Would it be possible, as a "workaround" to have "system-CC" and
> "ports-CC" defined in make.conf, making one CC the compiler for /usr/src
> and another for ports, or would this just create debugging nightmares?
>
> //Svein

If setup properly, something similar to what Gentoo Linux for profiled
compilers would be a very nice thing to have for `muxing' between
development tools. It would just make upgrades potentially more
painful though, and it would make tracking viral license tainting more
difficult for folks who are on the receiving end, or ones that aren't
paying attention to the licensing in the pieces of software they're
distributing...

The problem is that gcc is indeed lightyears ahead of any other
[opensource] compiler available, and the fact that there are a number
of bugs being filed against newer releases of gcc which don't
translate back necessarily into our version, cleanly. Similarly, gdb
has a lot of years of tread behind it as a debugger. If we were to
drop either the compiler or the debugger (and I'm pretty sure we'd
have to dump both if we dumped one), I believe that it would take
another group of individuals an extensive period of time to get back
to speed with what's available thanks to the GNU folks now... even if
the GNU stuff in its current state is buggy.

One thing I'm curious about though: for the bugs that do matter to us,
if there is a patch, would the patch be considered GPLv2 or GPLv3
licensed code? Not a lawyer, and I don't expect a lawyer's PoV, but if
the fix is truly trivial and it's compatible with the existing
license, that would be cool to import I would think...

Yes, we should really upgrade binutils though.. newer binutils would
bring in a number of bugfixes as well as feature enhancements...

Cheers,
-Garrett

PS Huzzah for R Stallman and his licensing crusade ;(...