Mailing List Archive

Hi!

I'm already use gcc-4.4.0/gcc-4.4.1 and whole system was rebuild with new gcc
(including kde-4.2.x/4.3.x and openoffice). Also i noticed that some scientific
software works better with gcc-4.4.x (like gromacs and gamess)

On óÒÅÄÁ 05 Á×ÇÕÓÔÁ 2009 05:15:57 Mark Loeser wrote:
> I'd really like to unmask gcc-4.4.1 soon, as in the next week or so.
> If you could please install it and test it out, I would appreciate it.
> Also, if you have any gcc 4.4 porting bugs assigned to a herd that you
> are a part of, resolving those bugs would help a lot.
>
> Thanks to all that have contributed and helped already,

--
Alexey 'Alexxy' Shvetsov
Gentoo/KDE
Gentoo/MIPS
Gentoo Team Ru

On Tue, 4 Aug 2009 21:15:57 -0400
Mark Loeser <halcy0n@gentoo.org> wrote:

> I'd really like to unmask gcc-4.4.1 soon, as in the next week or so.
> If you could please install it and test it out, I would appreciate it.
> Also, if you have any gcc 4.4 porting bugs assigned to a herd that you
> are a part of, resolving those bugs would help a lot.
>
> Thanks to all that have contributed and helped already,
>

Also the glibc-2.10 tracker, a lot of which are triggered by the combination
of it and gcc-4.4. I admit to massive slacking getting these fixed (mostly
due to my job sending me to the middle of nowhere for three weeks last month,
but also Persona 4 (oh that wacky Teddie)).

https://bugs.gentoo.org/showdependencytree.cgi?id=270353&hide_resolved=1

--
gcc-porting, Character is what you are in the dark.
treecleaner,
wxwidgets @ gentoo EFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662

Ryan Hill wrote:
> On Tue, 4 Aug 2009 21:15:57 -0400
> Mark Loeser <halcy0n@gentoo.org> wrote:
>
>> I'd really like to unmask gcc-4.4.1 soon, as in the next week or so.
>> If you could please install it and test it out, I would appreciate it.
>> Also, if you have any gcc 4.4 porting bugs assigned to a herd that you
>> are a part of, resolving those bugs would help a lot.
>>
>> Thanks to all that have contributed and helped already,
>>
>
> Also the glibc-2.10 tracker, a lot of which are triggered by the combination
> of it and gcc-4.4. I admit to massive slacking getting these fixed (mostly
> due to my job sending me to the middle of nowhere for three weeks last month,
> but also Persona 4 (oh that wacky Teddie)).
>
> https://bugs.gentoo.org/showdependencytree.cgi?id=270353&hide_resolved=1
>

I'd say what is left unfixed in both trackers gets a CC to treecleaners
in a week or so. Doesn't make sense to keep pkgs that won't build in tree.

-Samuli

On 08/05/2009 04:15 AM, Mark Loeser wrote:
> I'd really like to unmask gcc-4.4.1 soon, as in the next week or so.
> If you could please install it and test it out, I would appreciate it.
> Also, if you have any gcc 4.4 porting bugs assigned to a herd that you
> are a part of, resolving those bugs would help a lot.
>
> Thanks to all that have contributed and helped already,

I see that the "graphite" USE flag is disabled by default. Are there
any known issues with this new optimizer?

Nikos Chantziaras <realnc@arcor.de> said:
> I see that the "graphite" USE flag is disabled by default. Are there any
> known issues with this new optimizer?

None that I know of, but then again, I have not really played with it a
whole lot to say one way or the other. If you want to test it, please
file any bugs you find to toolchain.

Thanks,

--
Mark Loeser
email - halcy0n AT gentoo DOT org
email - mark AT halcy0n DOT com
web - http://www.halcy0n.com

>>>>> "Nikos" == Nikos Chantziaras <realnc@arcor.de> writes:

Nikos> I see that the "graphite" USE flag is disabled by default.
Nikos> Are there any known issues with this new optimizer?

I haven't found any using it on march=pentium3.

It looks like debian has it enabled in their gcc-4.4 packages, and
I've not found any problems on deb boxen (32 and 64 bit systems on
modern chips) either.

The USE flag probably ought to be +.

-JimC
--
James Cloos <cloos@jhcloos.com> OpenPGP: 1024D/ED7DAEA6

James Cloos wrote:
> Nikos> I see that the "graphite" USE flag is disabled by default.
> Nikos> Are there any known issues with this new optimizer?
> It looks like debian has it enabled in their gcc-4.4 packages, and
> I've not found any problems on deb boxen (32 and 64 bit systems on
> modern chips) either.
> The USE flag probably ought to be +.

I'd like to remind that we can't add EAPI's to system pkgs since it
would break the upgrading path.

If it works well, it should likely go in without USE flag...

-Samuli

On 08/13/2009 12:39 AM, Samuli Suominen wrote:
> James Cloos wrote:
>> Nikos> I see that the "graphite" USE flag is disabled by default.
>> Nikos> Are there any known issues with this new optimizer?
>> It looks like debian has it enabled in their gcc-4.4 packages, and
>> I've not found any problems on deb boxen (32 and 64 bit systems on
>> modern chips) either.
>> The USE flag probably ought to be +.
>
> I'd like to remind that we can't add EAPI's to system pkgs since it
> would break the upgrading path.
>
> If it works well, it should likely go in without USE flag...

Or a negating USE flag ("nographite" for example.)

On Thu, 13 Aug 2009 00:39:11 +0300
Samuli Suominen <ssuominen@gentoo.org> wrote:

> James Cloos wrote:
> > Nikos> I see that the "graphite" USE flag is disabled by default.
> > Nikos> Are there any known issues with this new optimizer?
> > It looks like debian has it enabled in their gcc-4.4 packages, and
> > I've not found any problems on deb boxen (32 and 64 bit systems on
> > modern chips) either.
> > The USE flag probably ought to be +.
>
> I'd like to remind that we can't add EAPI's to system pkgs since it
> would break the upgrading path.
>
> If it works well, it should likely go in without USE flag...

It's best to keep the number of external packages needed to build the
toolchain to a bare minimum. Anything that isn't a hard requirement should
be a USE flag. We also need to be able to bootstrap without a C++ compiler,
which one of graphite's dependencies (PPL) needs.


--
gcc-porting, Character is what you are in the dark.
treecleaner,
wxwidgets @ gentoo EFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662

Ryan Hill wrote:
> It's best to keep the number of external packages needed to build the
> toolchain to a bare minimum. Anything that isn't a hard requirement should
> be a USE flag. We also need to be able to bootstrap without a C++ compiler,
> which one of graphite's dependencies (PPL) needs.

True. I should have checked that it pulls in external deps. The upgrade
path is gone anyway since we don't have EAPI=0 dev-lang/python anymore
in tree.

If we don't care about upgrade path anymore might as well get rid of
profiles/obsolete too? And I do care about it, but it's gone now.

-Samuli

On Thu, 13 Aug 2009 10:27:26 +0300
Samuli Suominen <ssuominen@gentoo.org> wrote:

> Ryan Hill wrote:
> > It's best to keep the number of external packages needed to build the
> > toolchain to a bare minimum. Anything that isn't a hard requirement should
> > be a USE flag. We also need to be able to bootstrap without a C++ compiler,
> > which one of graphite's dependencies (PPL) needs.
>
> True. I should have checked that it pulls in external deps. The upgrade
> path is gone anyway since we don't have EAPI=0 dev-lang/python anymore
> in tree.

Wow, I'm getting extremely tired of people pulling things like this.

> If we don't care about upgrade path anymore might as well get rid of
> profiles/obsolete too? And I do care about it, but it's gone now.


--
gcc-porting, Character is what you are in the dark.
treecleaner,
wxwidgets @ gentoo EFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662