[gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 1215 bytes --]

Hey everyone,

If anybody has good reasons why -r4 shouldn't be unleashed upon the
unsuspecting ~arch users, respond to this (on-list please) before 0000
UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
to deal with the expected deluge of reports.

Why is -r4 interesting again?
http://article.gmane.org/gmane.linux.gentoo.devel/22986 details a number
of the changes. I recently added a new patchset with some more,
including fixes to:

* keyboard model default and key repeating -- might fix a few "some of
my keys don't work" problems, because the default used to be pc101
rather than pc105 with the new kbd driver.
* libGL segfaults
* some xmodmap problems
* latest Xpm security fix allows absolute paths again (fixes the GIMP's
Xpm plugin)

We need more testers to see which other packages are broken regarding
how they test for X libraries. Some of them assume the libraries won't
be in /usr/lib so supply a -L/path/to/X/libs even if they're
in /usr/lib. This results in a naked -L, which breaks things. If anyone
sees things suddenly breaking, check config.log for this. See
http://bugs.gentoo.org/show_bug.cgi?id=72124 for an example fix.

Thanks,
Donnie

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 1176 bytes --]

maillog: 06/12/2004-02:48:55(-0800): Donnie Berkholz types
> Hey everyone,
> 
> If anybody has good reasons why -r4 shouldn't be unleashed upon the
> unsuspecting ~arch users, respond to this (on-list please) before 0000
> UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> to deal with the expected deluge of reports.

I gave -r4 a shot yesterday on my amd64. A few problem I encountered:

- xterm could not compile properly and was giving me the "locale not
  supported; could not open input method" errors. I haven't looked into
  the problem in detail
- the default xorg.conf that came with it had
  `FontPath "/usr/lib64/fonts/..."' which I belive has something to do
  with the default "/usr/X11R6/lib/X11/fonts/..." in -r3, which has to
  be "/usr/share/fonts" anyway.

Other than that it worked straight out of the box. Not sure why the xkb
directory was also moved to /usr/lib64/xkb though.

-- 
\/   Georgi Georgiev   \/ * UNIX is a Trademark of Bell                \/
/\    chutz@gg3.net    /\ Laboratories.                                /\
\/  +81(90)6266-1163   \/                                              \/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 1551 bytes --]

On Tue, 2004-12-07 at 00:42 +0900, Georgi Georgiev wrote:
> maillog: 06/12/2004-02:48:55(-0800): Donnie Berkholz types
> > Hey everyone,
> > 
> > If anybody has good reasons why -r4 shouldn't be unleashed upon the
> > unsuspecting ~arch users, respond to this (on-list please) before 0000
> > UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> > to deal with the expected deluge of reports.
> 
> I gave -r4 a shot yesterday on my amd64. A few problem I encountered:
> 
> - xterm could not compile properly and was giving me the "locale not
>   supported; could not open input method" errors. I haven't looked into
>   the problem in detail

No idea -- I just compiled xterm-196-r1 yesterday with it.

> - the default xorg.conf that came with it had
>   `FontPath "/usr/lib64/fonts/..."' which I belive has something to do
>   with the default "/usr/X11R6/lib/X11/fonts/..." in -r3, which has to
>   be "/usr/share/fonts" anyway.

Any idea why this wouldn't catch it?

    # Fix default config files after installing fonts
to /usr/share/fonts
    sed -i -e "s:/usr/X11R6/$(get_libdir)/X11/fonts:/usr/share/fonts:g"
\
        -e "s:/usr/$(get_libdir)/X11/fonts:/usr/share/fonts:g" \
        ${D}/etc/X11/xorg.conf.example

> 
> Other than that it worked straight out of the box. Not sure why the xkb
> directory was also moved to /usr/lib64/xkb though.

Everything in /usr/X11R6/lib moved. /usr/X11R6/lib is now a symlink.
Therefore, xkb moved along with it. Am I misunderstanding something?

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Hans-Werner Hilse]

Hi,

On Mon, 06 Dec 2004 10:17:55 -0800
Donnie Berkholz <spyderous@gentoo.org> wrote:

> > - the default xorg.conf that came with it had
> >   `FontPath "/usr/lib64/fonts/..."' which I belive has something to do
> >   with the default "/usr/X11R6/lib/X11/fonts/..." in -r3, which has to
> >   be "/usr/share/fonts" anyway.
> 
> Any idea why this wouldn't catch it?
> 
>     # Fix default config files after installing fonts
> to /usr/share/fonts
>     sed -i -e "s:/usr/X11R6/$(get_libdir)/X11/fonts:/usr/share/fonts:g"
> \
>         -e "s:/usr/$(get_libdir)/X11/fonts:/usr/share/fonts:g" \
>         ${D}/etc/X11/xorg.conf.example

I think because xorg installs things that belong in /usr/lib(64)/X11 into /usr/lib(64). (or is xkb really supposed to be in /usr/lib/xkb?). So this just can't match. The error must be somewhere else.

I think this little piece of host_def_setup is causing some of this (if I'm right, it's used for the fonts, too):
---snip
 # /usr/X11R6/lib/X11
                echo "#define UsrLibDir /usr/$(get_libdir)" >> ${HOSTCONF}
---snap

HWH

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 930 bytes --]

maillog: 06/12/2004-10:17:55(-0800): Donnie Berkholz types
> > Other than that it worked straight out of the box. Not sure why the xkb
> > directory was also moved to /usr/lib64/xkb though.
> 
> Everything in /usr/X11R6/lib moved. /usr/X11R6/lib is now a symlink.
> Therefore, xkb moved along with it. Am I misunderstanding something?

My point is that /usr/X11R6/lib/X11/xkb moved to /usr/lib/xkb,
/usr/X11R6/lib/X11/modules moved to /usr/lib/modules... If xkb and
modules had moved to /usr/lib/X11/xkb and /usr/lib/X11/modules it would
have made more sense, but their current location is a bit confusing to
me.

-- 
 )   Georgi Georgiev    ) Tobacco is a filthy weed, That from the       )
(     chutz@gg3.net    (  devil does proceed; It drains your purse,    (
 )  +81(90)6266-1163    ) it burns your clothes, And makes a chimney    )
(  ------------------- (  of your nose. -- B. Waterhouse               (

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 1929 bytes --]

maillog: 07/12/2004-06:37:03(+0900): Georgi Georgiev types
> maillog: 06/12/2004-10:17:55(-0800): Donnie Berkholz types
> > > Other than that it worked straight out of the box. Not sure why the xkb
> > > directory was also moved to /usr/lib64/xkb though.
> > 
> > Everything in /usr/X11R6/lib moved. /usr/X11R6/lib is now a symlink.
> > Therefore, xkb moved along with it. Am I misunderstanding something?
> 
> My point is that /usr/X11R6/lib/X11/xkb moved to /usr/lib/xkb,
> /usr/X11R6/lib/X11/modules moved to /usr/lib/modules... If xkb and
> modules had moved to /usr/lib/X11/xkb and /usr/lib/X11/modules it would
> have made more sense, but their current location is a bit confusing to
> me.

OK, in more detail:

- I now have stuff installed in
  /usr/lib64/{config,etc,getconfig,modules,xkb} ... for some reasons
  that I cannot pinpoint and/or articulate, these bother me.
- Without the "font-server" use flag (correct me if I'm wrong), ttmkfdir
  is not installed; Do I really not need it even for a no-font-server
  installation?
- No idea why, but /usr/X11R6/lib64/X11/xkb is now a symlink that points
  to ../../lib/xkb, which means that it essentially points to itself
  (/usr/X11R6/lib points to lib64).  This could be related to the fact
  that I now installed after first removing /usr/X11R6 ( "emerge -C
  xorg-x11" and all packages that owned /usr/X11R6 save for baselayout
  and groff, and then move /usr/X11R6 away), but the binary package I
  built on a system that had xorg -r3.

I'm off to school in an hour, so I'll leave -r4 compiling on a clean (no
/usr/X11R6) system to see if at least the xkb symlink starts pointing to
a proper location.

-- 
\    Georgi Georgiev   \  Remember, Grasshopper, falling down 1000     \
 /    chutz@gg3.net     / stairs begins by tripping over the first      /
\   +81(90)6266-1163   \  one. -- Confusion                            \

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 601 bytes --]

maillog: 07/12/2004-07:06:03(+0900): Georgi Georgiev types
> - Without the "font-server" use flag (correct me if I'm wrong), ttmkfdir
>   is not installed; Do I really not need it even for a no-font-server
>   installation?

It seems that because I was installing from a binary package, portage
didn't understand that I also need x11-misc/ttmkfdir for xorg-x11.

-- 
\/   Georgi Georgiev   \/ "Rome wasn't burned in a day. "              \/
/\    chutz@gg3.net    /\                                              /\
\/  +81(90)6266-1163   \/                                              \/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 598 bytes --]

On Tue, 2004-12-07 at 07:17 +0900, Georgi Georgiev wrote:
> maillog: 07/12/2004-07:06:03(+0900): Georgi Georgiev types
> > - Without the "font-server" use flag (correct me if I'm wrong), ttmkfdir
> >   is not installed; Do I really not need it even for a no-font-server
> >   installation?
> 
> It seems that because I was installing from a binary package, portage
> didn't understand that I also need x11-misc/ttmkfdir for xorg-x11.

Good catch. It appears that the (R)DEPEND cleanup a while back let
ttmkfdir slip out of RDEPEND, when in fact it is required to properly
set up fonts.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 2306 bytes --]

On Tue, 2004-12-07 at 07:06 +0900, Georgi Georgiev wrote:
> maillog: 07/12/2004-06:37:03(+0900): Georgi Georgiev types
> > maillog: 06/12/2004-10:17:55(-0800): Donnie Berkholz types
> > > > Other than that it worked straight out of the box. Not sure why the xkb
> > > > directory was also moved to /usr/lib64/xkb though.
> > > 
> > > Everything in /usr/X11R6/lib moved. /usr/X11R6/lib is now a symlink.
> > > Therefore, xkb moved along with it. Am I misunderstanding something?
> > 
> > My point is that /usr/X11R6/lib/X11/xkb moved to /usr/lib/xkb,
> > /usr/X11R6/lib/X11/modules moved to /usr/lib/modules... If xkb and
> > modules had moved to /usr/lib/X11/xkb and /usr/lib/X11/modules it would
> > have made more sense, but their current location is a bit confusing to
> > me.

Ahh, now I see. You want /usr/libdir/X11, not /usr/libdir for that
stuff. I looked into it a little, and that makes a lot of sense. It
seems to fit fairly well with the upstream intent, too.

Looking in the imake config files, X11.tmpl has this:
#ifndef LibDir
# ifdef ProjectRoot
#  define LibDir Concat(ProjectRoot,/lib/X11)
# else
#  define LibDir /usr/lib/X11
# endif
#endif

So, that seems pretty fair. I'll fix up the ebuild a bit to work with
that and commit it.

> 
> OK, in more detail:
> 
> - I now have stuff installed in
>   /usr/lib64/{config,etc,getconfig,modules,xkb} ... for some reasons
>   that I cannot pinpoint and/or articulate, these bother me.
> - Without the "font-server" use flag (correct me if I'm wrong), ttmkfdir
>   is not installed; Do I really not need it even for a no-font-server
>   installation?
> - No idea why, but /usr/X11R6/lib64/X11/xkb is now a symlink that points
>   to ../../lib/xkb, which means that it essentially points to itself
>   (/usr/X11R6/lib points to lib64).  This could be related to the fact
>   that I now installed after first removing /usr/X11R6 ( "emerge -C
>   xorg-x11" and all packages that owned /usr/X11R6 save for baselayout
>   and groff, and then move /usr/X11R6 away), but the binary package I
>   built on a system that had xorg -r3.

I just committed a fix for one hardcoded "lib" xkb symlink that slipped
in instead of get_libdir(), which should treat lib64 properly on the
profiles that use lib64.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 1151 bytes --]

maillog: 06/12/2004-16:38:38(-0800): Donnie Berkholz types
> > - No idea why, but /usr/X11R6/lib64/X11/xkb is now a symlink that points
> >   to ../../lib/xkb, which means that it essentially points to itself
> >   (/usr/X11R6/lib points to lib64).  This could be related to the fact
> >   that I now installed after first removing /usr/X11R6 ( "emerge -C
> >   xorg-x11" and all packages that owned /usr/X11R6 save for baselayout
> >   and groff, and then move /usr/X11R6 away), but the binary package I
> >   built on a system that had xorg -r3.
> 
> I just committed a fix for one hardcoded "lib" xkb symlink that slipped
> in instead of get_libdir(), which should treat lib64 properly on the
> profiles that use lib64.

Actually, my profile does not use lib64, but why does X11R6? 

$ readlink /etc/make.profile
../usr/portage/profiles/default-linux/amd64/2004.3
$ portageq envvar CONF_LIBDIR
lib

-- 
 )   Georgi Georgiev    ) "There is no statute of limitations on        )
(     chutz@gg3.net    (  stupidity." -- Randomly produced by a        (
 )  +81(90)6266-1163    ) computer program called Markov3.              )

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 890 bytes --]

maillog: 06/12/2004-02:48:55(-0800): Donnie Berkholz types
> Hey everyone,
> 
> If anybody has good reasons why -r4 shouldn't be unleashed upon the
> unsuspecting ~arch users, respond to this (on-list please) before 0000
> UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> to deal with the expected deluge of reports.

Here is another one:

is_kernel() does not work for me, probably because I use KERNEL_OUTPUT.
Since I had to move the ebuild to my overlay anyway (to add ~amd64) I
also changed it to inherit linux-info and use kernel_is() instead of
is_kernel(). Is there any reason not to use the linux-info eclass?

-- 
-*   Georgi Georgiev   -* Govern a great nation as you would cook a    -*
*-    chutz@gg3.net    *- small fish. Don't overdo it. -- Lao Tsu      *-
-*  +81(90)6266-1163   -*                                              -*

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

[gentoo-dev] HOSTCONF in global scope

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 753 bytes --]

maillog: 06/12/2004-02:48:55(-0800): Donnie Berkholz types
> Hey everyone,
> 
> If anybody has good reasons why -r4 shouldn't be unleashed upon the
> unsuspecting ~arch users, respond to this (on-list please) before 0000
> UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> to deal with the expected deluge of reports.

cosmetic:

I see that HOSTCONF is defined in the global scope, and also at the
beginning of host_def_setup(). The global scope definition is
unnecessary I believe.

-- 
\/   Georgi Georgiev   \/ In Pierre Trudeau, Canada has finally        \/
/\    chutz@gg3.net    /\ produced a Prime Minister worthy of          /\
\/  +81(90)6266-1163   \/ assassination. -- John Diefenbaker           \/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

[gentoo-dev] Double stripping in strip_execs()

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 1172 bytes --]

maillog: 06/12/2004-02:48:55(-0800): Donnie Berkholz types
> Hey everyone,
> 
> If anybody has good reasons why -r4 shouldn't be unleashed upon the
> unsuspecting ~arch users, respond to this (on-list please) before 0000
> UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> to deal with the expected deluge of reports.

Cosmetic:

Unless I am misreading something, the following will strip
${D}/usr/$(get_libdir)/*.so.* twice

        # Now do the libraries ...
        for x in ${D}/usr/{$(get_libdir),$(get_libdir)/opengl/${PN}/$(get_libdir)}/*.so.* \
            ${D}/usr/{$(get_libdir),$(get_libdir)/locale/$(get_libdir)/common}/*.so.*; do
            if [ -f ${x} ]; then
                echo "$(echo ${x/${D}})"
                ${STRIP} --strip-debug ${x} || :
            fi
        done

I know this will soon become $(get_libdir)/X11 or something, but thought
I'd mention it anyway.

-- 
()   Georgi Georgiev   () <elmo> unclean: err, the admin team do not   ()
()    chutz@gg3.net    () control the archive, that's the ftp cabal    ()
()  +81(90)6266-1163   () <elmo> get your cabals right, damn it :-P    ()

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 675 bytes --]

On Mon, 2004-12-06 at 16:38 -0800, Donnie Berkholz wrote:
> So, that seems pretty fair. I'll fix up the ebuild a bit to work with
> that and commit it.

Committed. You'll find that a bunch of stuff now ends up
in /usr/lib/X11, although there are a couple of directories directly
in /usr/lib. I think I caught all the related symlink paths and so
forth, but let me know if you find anything.

supernova lib # find ./ -type d -maxdepth 1 -mindepth 1
./X11
./modules
./pkgconfig
./Server -- USE=sdk
./opengl
supernova lib # find X11/ -type d -maxdepth 1 -mindepth 1
X11/config
X11/locale
X11/x11perfcomp
X11/xedit
X11/xkb
X11/etc
X11/getconfig
X11/fonts

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 1260 bytes --]

On Tue, 2004-12-07 at 10:01 +0900, Georgi Georgiev wrote:
> maillog: 06/12/2004-16:38:38(-0800): Donnie Berkholz types
> > > - No idea why, but /usr/X11R6/lib64/X11/xkb is now a symlink that points
> > >   to ../../lib/xkb, which means that it essentially points to itself
> > >   (/usr/X11R6/lib points to lib64).  This could be related to the fact
> > >   that I now installed after first removing /usr/X11R6 ( "emerge -C
> > >   xorg-x11" and all packages that owned /usr/X11R6 save for baselayout
> > >   and groff, and then move /usr/X11R6 away), but the binary package I
> > >   built on a system that had xorg -r3.
> > 
> > I just committed a fix for one hardcoded "lib" xkb symlink that slipped
> > in instead of get_libdir(), which should treat lib64 properly on the
> > profiles that use lib64.
> 
> Actually, my profile does not use lib64, but why does X11R6? 

Ah, right. Check out this:

    # on amd64 we need /usr/lib64/X11/locale/lib to be a symlink
    # created by the emul lib ebuild in order for adobe acrobat,
staroffice,
    # and a few other apps to work.
    use amd64 && get_libdir_override lib64
    # lib32 isnt a supported configuration (yet?)
    [ "${CONF_LIBDIR}" == "lib32" ] && get_libdir_override lib

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch -- ~sparc OK

[Ferris McCormick]

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

On Mon, 6 Dec 2004, Donnie Berkholz wrote:

> Hey everyone,
>
> If anybody has good reasons why -r4 shouldn't be unleashed upon the
> unsuspecting ~arch users, respond to this (on-list please) before 0000
> UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> to deal with the expected deluge of reports.
>

Sparc should be OK; it's been working fine for me for what seems
like forever.  I'm rebuilding; with luck we'll be able to get rid
of that DeprecatedKeboard-if-kernel<2.6 mess.

NOTE, esp. sparc people, some patches required for USE=dlloader
and so also USE='dlloader hardened' --- read and apply patches
from Bug 61063.

Also, sparc only, if you haven't seen this note from Donnie, please
make sure to read the rest of it and the extensive xorg-x11 ChangeLog
before proceeding.

> Why is -r4 interesting again?
> http://article.gmane.org/gmane.linux.gentoo.devel/22986 details a number
> of the changes. I recently added a new patchset with some more,
> including fixes to:
>
> * keyboard model default and key repeating -- might fix a few "some of
> my keys don't work" problems, because the default used to be pc101
> rather than pc105 with the new kbd driver.
> * libGL segfaults
> * some xmodmap problems
> * latest Xpm security fix allows absolute paths again (fixes the GIMP's
> Xpm plugin)
>
> We need more testers to see which other packages are broken regarding
> how they test for X libraries. Some of them assume the libraries won't
> be in /usr/lib so supply a -L/path/to/X/libs even if they're
> in /usr/lib. This results in a naked -L, which breaks things. If anyone
> sees things suddenly breaking, check config.log for this. See
> http://bugs.gentoo.org/show_bug.cgi?id=72124 for an example fix.
>
> Thanks,
> Donnie
>

Regards,
- --
Ferris McCormick (P44646, MI) <fmccor@gentoo.org>
Developer, Gentoo Linux (Sparc)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFBtaNXQa6M3+I///cRAk6PAKDRumB1Hee9k+YruJ2HOzgVmcCOtQCgnDN7
vfpIQOpOU7vcAFhfBBBpMR0=
=QzrO
-----END PGP SIGNATURE-----

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch --- ~sparc tests good.

[Ferris McCormick]

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

On Mon, 6 Dec 2004, Donnie Berkholz wrote:

> Hey everyone,
>
> If anybody has good reasons why -r4 shouldn't be unleashed upon the
> unsuspecting ~arch users, respond to this (on-list please) before 0000
> UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> to deal with the expected deluge of reports.
>

Sparc tests fine under the following conditions:

1. ebuild taken from cvs at 13:06 (GMT)
2. USE flags thus:
===========================
lacewing root # PKGDIR=${PD} emerge -pkv xorg-x11

These are the packages that I would merge, in order:

Calculating dependencies ...done!
[binary   R   ] x11-base/xorg-x11-6.8.0-r4  (-3dfx) (-3dnow) +bitmap-fonts 
- -cjk -debug +dlloader -dmx -doc -font-server +hardened +insecure-drivers 
- -ipv6 -minimal (-mmx) +nls +opengl +pam -sdk (-sse) -static 
+truetype-fonts +type1-fonts (-uclibc) -xprint +xv [1]
===========================

3. Because of the +dlloader, necessarily the patches from Bug 61063 were
    applied.
4. This version of X is hardened.
5. The (deprecated) keyboard driver is still required. kbd shifts keys as
    it always has.

6. NOTE: The two 'is_kernel' tests produce error messages, but (for me,
    at any rate) ebuild operation is correct anyway.  I have not
    investigated further, but as I recall, the messages were:
    [: wrong number of arguments (for the check kernel-2.2.x)
    Command '*' not understood

Given this, sparc is fine. ---

7. For non-(hardened+dlloader) users (everyone but me, I think),
    extra patching should not be needed, although at some point, the
    61063 patches are required.

Nice job,
Ferris
- --
Ferris McCormick (P44646, MI) <fmccor@gentoo.org>
Developer, Gentoo Linux (Sparc)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFBtguNQa6M3+I///cRAm7PAJ9uagni1SWVX7W4qXd4nkTFiRrYrgCfd2O7
xtPzhkVR3UqNF+Jk8P5jAZ8=
=1Uo0
-----END PGP SIGNATURE-----

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch --- ~sparc tests good.

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 3220 bytes --]

maillog: 07/12/2004-19:59:05(+0000): Ferris McCormick types
> On Mon, 6 Dec 2004, Donnie Berkholz wrote:
> 
> > Hey everyone,
> >
> > If anybody has good reasons why -r4 shouldn't be unleashed upon the
> > unsuspecting ~arch users, respond to this (on-list please) before 0000
> > UTC Wednesday. Why then? Well, my finals will be over, so I'll have time
> > to deal with the expected deluge of reports.
> >
> 
> Sparc tests fine under the following conditions:
> 
> 1. ebuild taken from cvs at 13:06 (GMT)
> 2. USE flags thus:
> ===========================
> lacewing root # PKGDIR=${PD} emerge -pkv xorg-x11
> 
> These are the packages that I would merge, in order:
> 
> Calculating dependencies ...done!
> [binary   R   ] x11-base/xorg-x11-6.8.0-r4  (-3dfx) (-3dnow) +bitmap-fonts 
> -cjk -debug +dlloader -dmx -doc -font-server +hardened +insecure-drivers 
> -ipv6 -minimal (-mmx) +nls +opengl +pam -sdk (-sse) -static 
> +truetype-fonts +type1-fonts (-uclibc) -xprint +xv [1]
> ===========================
> 
> 3. Because of the +dlloader, necessarily the patches from Bug 61063 were
>     applied.
> 4. This version of X is hardened.
> 5. The (deprecated) keyboard driver is still required. kbd shifts keys as
>     it always has.
> 
> 6. NOTE: The two 'is_kernel' tests produce error messages, but (for me,
>     at any rate) ebuild operation is correct anyway.  I have not
>     investigated further, but as I recall, the messages were:
>     [: wrong number of arguments (for the check kernel-2.2.x)
>     Command '*' not understood

This is what I had in my overlay while testing the ebuild yesterday.
The linux-info.eclass kernel_is() should only return an exit code so its
output should not be looked at. I.e. those "[" and "$(" are not needed.

--- /usr/portage/x11-base/xorg-x11/xorg-x11-6.8.0-r4.ebuild	2004-12-07 13:23:15.000000000 +0900
+++ ./xorg-x11-6.8.0-r4.ebuild	2004-12-07 17:59:24.000000000 +0900
@@ -670,7 +674,7 @@
                fi
 
                if ( [ -e "${ROOT}/usr/src/linux" ] \
-                       && [ ! $(kernel_is "2" "2") ] ) \
+                       && ! ( kernel_is "2" "2" ) ) \
                        || [ "$(uname -r | cut -d. -f1,2)" != "2.2" ]; then
                        echo "#define HasLinuxInput YES" >> ${HOSTCONF}
                fi
@@ -794,7 +798,7 @@
                                echo "#define ModuleAsCmd CcCmd -c -x assembler -fno-pie -fno-PIE" >> ${HOSTCONF}
                        fi
                        if ( [ -e "${ROOT}/usr/src/linux" ] \
-                               && !( $(kernel_is "2" "6") ) ) \
+                               && !( kernel_is "2" "6" ) ) \
                                || [ "$(uname -r | cut -d. -f1,2)" != "2.6" ]; then
                                einfo "Building for kernels less than 2.6 requires special treatment"
                                echo "#define UseDeprecatedKeyboardDriver YES" >> ${HOSTCONF}


-- 
\    Georgi Georgiev   \  Don't lose Your head To gain a minute You    \
 /    chutz@gg3.net     / need your head Your brains are in it. --      /
\   +81(90)6266-1163   \  Burma Shave                                  \

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch --- ~sparc tests good.

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 955 bytes --]

maillog: 08/12/2004-07:52:19(+0900): Georgi Georgiev types
> maillog: 07/12/2004-19:59:05(+0000): Ferris McCormick types
> > 6. NOTE: The two 'is_kernel' tests produce error messages, but (for me,
> >     at any rate) ebuild operation is correct anyway.  I have not
> >     investigated further, but as I recall, the messages were:
> >     [: wrong number of arguments (for the check kernel-2.2.x)
> >     Command '*' not understood
> 
> This is what I had in my overlay while testing the ebuild yesterday.
> The linux-info.eclass kernel_is() should only return an exit code so its
> output should not be looked at. I.e. those "[" and "$(" are not needed.

Ah, well, linux-info broke in the meantime, that's all.

-- 
/    Georgi Georgiev   /  A year spent in artificial intelligence is   /
\     chutz@gg3.net    \  enough to make one believe in God.           \
/   +81(90)6266-1163   /                                               /

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch --- ~sparc tests good.

[Ferris McCormick]

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

On Wed, 8 Dec 2004, Georgi Georgiev wrote:

> maillog: 08/12/2004-07:52:19(+0900): Georgi Georgiev types
>> maillog: 07/12/2004-19:59:05(+0000): Ferris McCormick types
>>> 6. NOTE: The two 'is_kernel' tests produce error messages, but (for me,
>>>     at any rate) ebuild operation is correct anyway.  I have not
>>>     investigated further, but as I recall, the messages were:
>>>     [: wrong number of arguments (for the check kernel-2.2.x)
>>>     Command '*' not understood
>>
>> This is what I had in my overlay while testing the ebuild yesterday.
>> The linux-info.eclass kernel_is() should only return an exit code so its
>> output should not be looked at. I.e. those "[" and "$(" are not needed.
>
> Ah, well, linux-info broke in the meantime, that's all.
>

Thanks for the information.  For your confusion, there is a 'kernel_is'
in linux-info.eclass, and an 'is_kernel' in x11.eclass.  The ebuilds
used to use is_kernel; now they use kernel_is, and perhaps the
difficulties started with that?


Regards,
- --
Ferris McCormick (P44646, MI) <fmccor@gentoo.org>
Developer, Gentoo Linux (sparc)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFBtj6WQa6M3+I///cRAk6iAKCJYpG+ZXeJr0je3W7Wu8CmdPLK1QCgiy0y
/RfYqDogLRPD9rRMSgYcHxw=
=6A0o
-----END PGP SIGNATURE-----

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch --- ~sparc tests good.

[Georgi Georgiev]

maillog: 07/12/2004-23:36:49(+0000): Ferris McCormick types
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Wed, 8 Dec 2004, Georgi Georgiev wrote:
> 
> >maillog: 08/12/2004-07:52:19(+0900): Georgi Georgiev types
> >>maillog: 07/12/2004-19:59:05(+0000): Ferris McCormick types
> >>>6. NOTE: The two 'is_kernel' tests produce error messages, but (for me,
> >>>    at any rate) ebuild operation is correct anyway.  I have not
> >>>    investigated further, but as I recall, the messages were:
> >>>    [: wrong number of arguments (for the check kernel-2.2.x)
> >>>    Command '*' not understood
> >>
> >>This is what I had in my overlay while testing the ebuild yesterday.
> >>The linux-info.eclass kernel_is() should only return an exit code so its
> >>output should not be looked at. I.e. those "[" and "$(" are not needed.
> >
> >Ah, well, linux-info broke in the meantime, that's all.
> >
> 
> Thanks for the information.  For your confusion, there is a 'kernel_is'
> in linux-info.eclass, and an 'is_kernel' in x11.eclass.  The ebuilds
> used to use is_kernel; now they use kernel_is, and perhaps the
> difficulties started with that?

is_kernel will probably dissapear soon enough.

And to the attention of spyderous: kernel_is from linux-info returns a
valid exit code and outputs lots of einfos, therefore the [ $(kernel_is
..) ] and the other $(kernel_is) must not be in a subshell whose output
is passed to the command line (hence the "Command '*' not understood").

And lastly, well, sorry for the misinformation. linux-info did indeed
break, but only for me I guess (am I the only one using KERNEL_OUTPUT,
because nobody else seems to complain when stuff related to
KERNEL_OUTPUT is obviously broken), which was unrelated to the current
problem. It's just that when I tried it, there were two problems to
solve.

-- 
\/   Georgi Georgiev   \/ Straw? No, too stupid a fad. I put soot on   \/
/\    chutz@gg3.net    /\ warts.                                       /\
\/  +81(90)6266-1163   \/                                              \/

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch --- ~sparc tests good.

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 421 bytes --]

On Wed, 2004-12-08 at 09:16 +0900, Georgi Georgiev wrote:
> And to the attention of spyderous: kernel_is from linux-info returns a
> valid exit code and outputs lots of einfos, therefore the [ $(kernel_is
> ..) ] and the other $(kernel_is) must not be in a subshell whose output
> is passed to the command line (hence the "Command '*' not understood").

I'm accepting patches that retain the former functionality.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] xorg-x11-6.8.0-r4 to go ~arch --- ~sparc tests good.

[Georgi Georgiev]

[-- Attachment #1: Type: text/plain, Size: 2122 bytes --]

maillog: 07/12/2004-16:38:16(-0800): Donnie Berkholz types
> On Wed, 2004-12-08 at 09:16 +0900, Georgi Georgiev wrote:
> > And to the attention of spyderous: kernel_is from linux-info returns a
> > valid exit code and outputs lots of einfos, therefore the [ $(kernel_is
> > ..) ] and the other $(kernel_is) must not be in a subshell whose output
> > is passed to the command line (hence the "Command '*' not understood").
> 
> I'm accepting patches that retain the former functionality.

I posted it four levels up in this thread, but here it is again anyway:

--- /usr/portage/x11-base/xorg-x11/xorg-x11-6.8.0-r4.ebuild	2004-12-07 13:23:15.000000000 +0900
+++ ./xorg-x11-6.8.0-r4.ebuild	2004-12-07 17:59:24.000000000 +0900
@@ -670,7 +674,7 @@
                fi
 
                if ( [ -e "${ROOT}/usr/src/linux" ] \
-                       && [ ! $(kernel_is "2" "2") ] ) \
+                       && !( kernel_is "2" "2" ) ) \
                        || [ "$(uname -r | cut -d. -f1,2)" != "2.2" ]; then
                        echo "#define HasLinuxInput YES" >> ${HOSTCONF}
                fi
@@ -794,7 +798,7 @@
                                echo "#define ModuleAsCmd CcCmd -c -x assembler -fno-pie -fno-PIE" >> ${HOSTCONF}
                        fi
                        if ( [ -e "${ROOT}/usr/src/linux" ] \
-                               && !( $(kernel_is "2" "6") ) ) \
+                               && !( kernel_is "2" "6" ) ) \
                                || [ "$(uname -r | cut -d. -f1,2)" != "2.6" ]; then
                                einfo "Building for kernels less than 2.6 requires special treatment"
                                echo "#define UseDeprecatedKeyboardDriver YES" >> ${HOSTCONF}

It is a bit off, because I also have a patch for xft that makes it
generate bold fonts. You can still apply it manually anyway.

-- 
 )   Georgi Georgiev    ) The one charm of marriage is that it makes    )
(     chutz@gg3.net    (  a life of deception a neccessity. -- Oscar   (
 )  +81(90)6266-1163    ) Wilde                                         )

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]