[gentoo-dev] Optimized Qt

[gentoo-dev] Optimized Qt

[Prashanth Aditya Susarla]

Someone wanted to know if Qt compiled with higher optimization than its 
default was stable or not. I have just merged Qt compiled with 
-march=athlon-tbird -O3 -fomit-frame-pointer -pipe -mmmx -m3dnow (gcc-3.1, 
binutils-2.12.90.0.7, glibc-2.2.5-r4) and the emerge completed 
successfully. I am also working in my KDE now and there seem to be no 
breakages/crashes.
Untar qt-copy-3.0.4.tar.bz2 and edit 
qt-copy-3.0.4/mkspecs/linux-g++/qmake.conf (it was linux-g++ in my case, 
it could be different for others) and re-archive the entire tree. Then 
ebuild qt-3.0.4-20020606-r1.ebuild digest would recompute the digests and 
one can proceed with the emerge. Obviously, this new digest file would be 
overwritten in the next emerge rsync.

Regards,
Prashanth Aditya Susarla

Re: [gentoo-dev] Optimized Qt

[Alexander Holler]

Hello,

--On Freitag, Juni 07, 2002 11:50:38 +0530 Prashanth Aditya Susarla 
<aditya@iitk.ac.in> wrote:

> Someone wanted to know if Qt compiled with higher optimization than its
> default was stable or not. I have just merged Qt compiled with
> -march=athlon-tbird -O3 -fomit-frame-pointer -pipe -mmmx -m3dnow
> (gcc-3.1,  binutils-2.12.90.0.7, glibc-2.2.5-r4) and the emerge completed
> successfully. I am also working in my KDE now and there seem to be no
> breakages/crashes.
> Untar qt-copy-3.0.4.tar.bz2 and edit
> qt-copy-3.0.4/mkspecs/linux-g++/qmake.conf (it was linux-g++ in my case,
> it could be different for others) and re-archive the entire tree. Then
> ebuild qt-3.0.4-20020606-r1.ebuild digest would recompute the digests and
> one can proceed with the emerge. Obviously, this new digest file would be
> overwritten in the next emerge rsync.

Good idea, maybe this could go into the next ebuild, here's a patch for the 
actual ebuild:


---------------------------------------------------------------------------
----------------
--- qt-3.0.4.20020606-r1.ebuild.ori     2002-06-08 09:59:05.000000000 +0200
+++ qt-3.0.4.20020606-r1.ebuild 2002-06-08 10:22:09.000000000 +0200
@@ -49,6 +49,8 @@
        cp configure configure.orig
        sed -e 's:read acceptance:acceptance=yes:' \
            -e 's:|-repeater|:|-nas-sound|-repeater|:' configure.orig > 
configure
+       cp mkspecs/linux-g++/qmake.conf mkspecs/linux-g++/qmake.conf.orig
+       sed -e "s/-O2/${CFLAGS}/" mkspecs/linux-g++/qmake.conf.orig > 
mkspecs/linux-g++/qmake.conf

 }
---------------------------------------------------------------------------
----------------


Regards,

Alexander

Re: [gentoo-dev] Optimized Qt

[Dan Armak]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

Please see bug #1571. I haven't had time to "investigate" beyond what the 
comments say. If we could get a good rule of what CFLAGS work and what don't 
I'd be willing to itroduce the thing into the qt ebuilds.


On Saturday 08 June 2002 11:23, Alexander Holler wrote:
> Hello,
>
> --On Freitag, Juni 07, 2002 11:50:38 +0530 Prashanth Aditya Susarla
>
> <aditya@iitk.ac.in> wrote:
> > Someone wanted to know if Qt compiled with higher optimization than its
> > default was stable or not. I have just merged Qt compiled with
> > -march=athlon-tbird -O3 -fomit-frame-pointer -pipe -mmmx -m3dnow
> > (gcc-3.1,  binutils-2.12.90.0.7, glibc-2.2.5-r4) and the emerge completed
> > successfully. I am also working in my KDE now and there seem to be no
> > breakages/crashes.
> > Untar qt-copy-3.0.4.tar.bz2 and edit
> > qt-copy-3.0.4/mkspecs/linux-g++/qmake.conf (it was linux-g++ in my case,
> > it could be different for others) and re-archive the entire tree. Then
> > ebuild qt-3.0.4-20020606-r1.ebuild digest would recompute the digests and
> > one can proceed with the emerge. Obviously, this new digest file would be
> > overwritten in the next emerge rsync.
>
> Good idea, maybe this could go into the next ebuild, here's a patch for the
> actual ebuild:
>
>
> ---------------------------------------------------------------------------
> ----------------
> --- qt-3.0.4.20020606-r1.ebuild.ori     2002-06-08 09:59:05.000000000 +0200
> +++ qt-3.0.4.20020606-r1.ebuild 2002-06-08 10:22:09.000000000 +0200
> @@ -49,6 +49,8 @@
>         cp configure configure.orig
>         sed -e 's:read acceptance:acceptance=yes:' \
>             -e 's:|-repeater|:|-nas-sound|-repeater|:' configure.orig >
> configure
> +       cp mkspecs/linux-g++/qmake.conf mkspecs/linux-g++/qmake.conf.orig
> +       sed -e "s/-O2/${CFLAGS}/" mkspecs/linux-g++/qmake.conf.orig >
> mkspecs/linux-g++/qmake.conf
>
>  }
> ---------------------------------------------------------------------------
> ----------------
>
>
> Regards,
>
> Alexander
>
> _______________________________________________
> gentoo-dev mailing list
> gentoo-dev@gentoo.org
> http://lists.gentoo.org/mailman/listinfo/gentoo-dev

- -- 
Dan Armak
Gentoo Linux developer (KDE)
Matan, Israel
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE9AclJUI2RQ41fiVERAgtLAJ9cfL3CWk0li6TfoosljviGfbpwigCbBTei
DZYflJdfuVVy7o/eED5nkvA=
=l92L
-----END PGP SIGNATURE-----

Re: [gentoo-dev] Optimized Qt

[Alexander Holler]

Hello,

--On Samstag, Juni 08, 2002 12:07:17 +0300 Dan Armak <danarmak@gentoo.org> 
wrote:

> Please see bug #1571. I haven't had time to "investigate" beyond what the
> comments say. If we could get a good rule of what CFLAGS work and what
> don't  I'd be willing to itroduce the thing into the qt ebuilds.

I've just got the same bug with reloads building qt3 with customized 
CFLAGS. This gcc bug looks like some bugs I've got trying to build gentoo 
with gcc 3.1. With the difference that gcc 3.1 kindly asks for filling out 
bug reports. ;)

And I agree that it isn't a good idea to use customized flags if qmake 
records these flags for future buildings (specially the processor flags).

Alexander