Gtk+ 2.6.10 w/ newer Glibs

Gtk+ 2.6.10 w/ newer Glibs

[Michael 'Mickey' Lauer]

Since .dev upgraded gnome-vfs and removed gnome-vfs-dbus, gnome-vfs is
no longer buildable against an old (<= 2.9.4) glib. Gtk+ 2.6.10 (which is still
in wide use today) again is not buildable against newer glib variants.

Any glib / Gtk+ experts to fix the problem? The first issue out is
here:

|  arm-linux-gcc -march=armv4t -mtune=arm920t -DHAVE_CONFIG_H -I. -I/local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/gtk+-2.6.10/modules/engines/pixbuf -I../../.. -I/local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/gtk+-2.6.10 -I/local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/gtk+-2.6.10/gdk -I../../../gdk -DG_DISABLE_DEPRECATED -DGDK_PIXBUF_DISABLE_DEPRECATED -DGDK_DISABLE_DEPRECATED -pthread -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include/glib-2.0 -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/lib/glib-2.0/include -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include/pango-1.0 -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include/freetype2 -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include/atk-1.0 -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include -I/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include -isystem/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include -isystem/local/pkg/openmoko/gta01/tmp/staging/arm-linux/include -fexpensive-optimizations -fomit-frame-pointer -frename-registers -O2 -Wall -MT pixbuf-draw.lo -MD -MP -MF .deps/pixbuf-draw.Tpo -c pixbuf-draw.c  -fPIC -DPIC -o .libs/pixbuf-draw.o
| In file included from ../../../gtk/gtk.h:60,
|                  from pixbuf.h:23,
|                  from pixbuf-draw.c:28:
| ../../../gtk/gtkclist.h:155: error: expected specifier-qualifier-list before 'GMemChunk'
| In file included from ../../../gtk/gtk.h:152,
|                  from pixbuf.h:23,
|                  from pixbuf-draw.c:28:
| ../../../gtk/gtkstatusbar.h:68: error: expected specifier-qualifier-list before 'GMemChunk'
| make[4]: *** [pixbuf-draw.lo] Fehler 1
| make[4]: Leaving directory `/local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/gtk+-2.6.10/modules/engines/pixbuf'
| make[3]: *** [all-recursive] Fehler 1
| make[3]: Leaving directory `/local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/gtk+-2.6.10/modules/engines'
| make[2]: *** [all-recursive] Fehler 1
| make[2]: Leaving directory `/local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/gtk+-2.6.10/modules'
| make[1]: *** [all-recursive] Fehler 1
| make[1]: Leaving directory `/local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/gtk+-2.6.10'
| make: *** [all] Fehler 2
| FATAL: oe_runmake failed
NOTE: Task failed: /local/pkg/openmoko/gta01/tmp/work/armv4t-linux/gtk+-2.6.10-r4/temp/log.do_compile.30492
NOTE: package gtk+-2.6.10-r4: task do_compile: failed
ERROR: TaskFailed event exception, aborting
NOTE: package gtk+-2.6.10: failed
ERROR: Build of 'gtk+' failed

Regards,

:M:
-- 
Michael 'Mickey' Lauer | IT-Freelancer | http://www.vanille-media.de

Re: Gtk+ 2.6.10 w/ newer Glibs

[Koen Kooi]

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

Michael 'Mickey' Lauer schreef:
> Since .dev upgraded gnome-vfs and removed gnome-vfs-dbus, gnome-vfs is
> no longer buildable against an old (<= 2.9.4) glib. Gtk+ 2.6.10 (which is still
> in wide use today) again is not buildable against newer glib variants.
> 
> Any glib / Gtk+ experts to fix the problem? The first issue out is
> here:

> | ../../../gtk/gtkclist.h:155: error: expected specifier-qualifier-list before 'GMemChunk'

I think this is caused by the work done on g_slice. You could have a look at backporting
those changes[1].

regards,

Koen


[1] Or use the maemo patches that do that :)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFFp+7SMkyGM64RGpERAiKrAJ9QEzFmgcYDZ3VHjKmRtLYjhTcAfgCfa9z2
IrzECKchJ+PAsQuiz9knmnI=
=fkQf
-----END PGP SIGNATURE-----

Re: Gtk+ 2.6.10 w/ newer Glibs

[Michael 'Mickey' Lauer]

Koen Kooi wrote:
> Michael 'Mickey' Lauer schreef:
>> Since .dev upgraded gnome-vfs and removed gnome-vfs-dbus, gnome-vfs is
>> no longer buildable against an old (<= 2.9.4) glib. Gtk+ 2.6.10 (which is still
>> in wide use today) again is not buildable against newer glib variants.
>> 
>> Any glib / Gtk+ experts to fix the problem? The first issue out is
>> here:

>> | ../../../gtk/gtkclist.h:155: error: expected specifier-qualifier-list before 'GMemChunk'

> I think this is caused by the work done on g_slice. You could have a look at backporting
> those changes[1].
[...]
> [1] Or use the maemo patches that do that :)

Heh, I'd like to, but how am I supposed to do that.
They don't seem to maintain patchsets or do they?

I'm afraid digging the proper changes out of their dozens of tags and
branches is way over my capabilities.

Regards,

:M:
-- 
Michael 'Mickey' Lauer | IT-Freelancer | http://www.vanille-media.de

Re: Gtk+ 2.6.10 w/ newer Glibs

[Richard Purdie]

On Fri, 2007-01-12 at 20:43 +0100, Michael 'Mickey' Lauer wrote:
> Since .dev upgraded gnome-vfs and removed gnome-vfs-dbus, gnome-vfs is
> no longer buildable against an old (<= 2.9.4) glib. Gtk+ 2.6.10 (which is still
> in wide use today) again is not buildable against newer glib variants.
> 
> Any glib / Gtk+ experts to fix the problem? The first issue out is
> here:

I asked someone:

> |  arm-linux-gcc ... -DG_DISABLE_DEPRECATED

G_DISABLE_DEPRECATED tells GLib to hide all deprecated functions.  They
still exist in the library, but they are hidden in the headers.  It's
considered bad form to do this in release code, because sometimes people
want to run older apps against newer libraries.

> | ../../../gtk/gtkstatusbar.h:68: error: expected
specifier-qualifier-list before 'GMemChunk'

GMemChunk was deprecated in GLib 2.10.

To fix this there needs to be a patch that removes
-DG_DISABLE_DEPRECATED from every Makefile.am.

Cheers,

Richard

Re: Gtk+ 2.6.10 w/ newer Glibs

[Koen Kooi]

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

Richard Purdie schreef:

> To fix this there needs to be a patch that removes
> -DG_DISABLE_DEPRECATED from every Makefile.am.

patch? sed! :)

regards,

Koen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFFqQWnMkyGM64RGpERAhWQAJ0Yz8MgM8LI685bY6OeLc98/KIf/QCeLIa3
4lvbv2nlpod5CwW72zmvJrc=
=iWJb
-----END PGP SIGNATURE-----

Re: Angstrom cannot buile portabase

[Holger Freyther]

Am 13.01.2007 um 23:39 schrieb Alban J Pearce:

> I have installed Angstrom
> bitbake , nano works
>
>  NOTE: build 200701132219: started
>
> OE Build Configuration:
> BB_VERSION     = "1.6.2"
> OE_REVISION    = "28548b0eda864e1b3a44136bcced77ee8c1e9dab"
> TARGET_ARCH    = "arm"
> TARGET_OS      = "linux-gnueabi"
> MACHINE        = "spitz"
> DISTRO         = "angstrom"
> DISTRO_VERSION = "test-20070113"
> TARGET_FPU     = "soft"
>
> But bitbake portsabase fails
>
> NOTE: Applying patch 'kernel-keymap.patch'
> Command Error: exit status: 1  Output:
> Applying patch kernel-keymap.patch
> can't find file to patch at input line 10
> Perhaps you used the wrong -p or --strip option?
> The text leading up to this was:


Hi,

did you run out of disk space?

h.

Re: Angstrom cannot buile portabase

[Koen Kooi]

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

Holger Freyther schreef:
> Am 13.01.2007 um 23:39 schrieb Alban J Pearce:
> 
>> I have installed Angstrom
>> bitbake , nano works
>>
>>  NOTE: build 200701132219: started
>>
>> OE Build Configuration:
>> BB_VERSION     = "1.6.2"
>> OE_REVISION    = "28548b0eda864e1b3a44136bcced77ee8c1e9dab"
>> TARGET_ARCH    = "arm"
>> TARGET_OS      = "linux-gnueabi"
>> MACHINE        = "spitz"
>> DISTRO         = "angstrom"
>> DISTRO_VERSION = "test-20070113"
>> TARGET_FPU     = "soft"
>>
>> But bitbake portsabase fails
>>
>> NOTE: Applying patch 'kernel-keymap.patch'
>> Command Error: exit status: 1  Output:
>> Applying patch kernel-keymap.patch
>> can't find file to patch at input line 10
>> Perhaps you used the wrong -p or --strip option?
>> The text leading up to this was:
> 
> 
> Hi,
> 
> did you run out of disk space?

This again highlights why angstrom doesn't support Opie. It is broken. Feel free to fix it.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFFqXV1MkyGM64RGpERAvzXAJ4ieApVu7oipWI09HadA85S8sXYhgCeKvfR
ZIlBSjqhdWBzzfmUmFBiLxs=
=W9Rb
-----END PGP SIGNATURE-----

Re: Angstrom cannot buile portabase

[GoXbox Live]

Just for information.
Opie is not broken. There are now several of us wich build Opie right out of
oe.dev, without any hacks.
So please stop this nonsense about "Opie doesnt work" / "Opie doesnt build"
and so on.
Paul Psokolovsky has realy done a great job with Opie in the oe.dev.

gxl

2007/1/14, Koen Kooi <koen@dominion.kabel.utwente.nl>:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Holger Freyther schreef:
> > Am 13.01.2007 um 23:39 schrieb Alban J Pearce:
> >
> >> I have installed Angstrom
> >> bitbake , nano works
> >>
> >>  NOTE: build 200701132219: started
> >>
> >> OE Build Configuration:
> >> BB_VERSION     = "1.6.2"
> >> OE_REVISION    = "28548b0eda864e1b3a44136bcced77ee8c1e9dab"
> >> TARGET_ARCH    = "arm"
> >> TARGET_OS      = "linux-gnueabi"
> >> MACHINE        = "spitz"
> >> DISTRO         = "angstrom"
> >> DISTRO_VERSION = "test-20070113"
> >> TARGET_FPU     = "soft"
> >>
> >> But bitbake portsabase fails
> >>
> >> NOTE: Applying patch 'kernel-keymap.patch'
> >> Command Error: exit status: 1  Output:
> >> Applying patch kernel-keymap.patch
> >> can't find file to patch at input line 10
> >> Perhaps you used the wrong -p or --strip option?
> >> The text leading up to this was:
> >
> >
> > Hi,
> >
> > did you run out of disk space?
>
> This again highlights why angstrom doesn't support Opie. It is broken.
> Feel free to fix it.
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.5 (Darwin)
>
> iD8DBQFFqXV1MkyGM64RGpERAvzXAJ4ieApVu7oipWI09HadA85S8sXYhgCeKvfR
> ZIlBSjqhdWBzzfmUmFBiLxs=
> =W9Rb
> -----END PGP SIGNATURE-----
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: Angstrom cannot buile portabase

[Koen Kooi]

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

GoXbox Live schreef:
> Just for information.
> Opie is not broken. There are now several of us wich build Opie right out of
> oe.dev, without any hacks.
> So please stop this nonsense about "Opie doesnt work" / "Opie doesnt build"
> and so on.

Yes, I can clearly see how well it works and builds when kernel-keymap.patch fails to apply.


> Paul Psokolovsky has realy done a great job with Opie in the oe.dev.
> 
> gxl
> 
> 2007/1/14, Koen Kooi <koen@dominion.kabel.utwente.nl>:
> Holger Freyther schreef:
>>>> Am 13.01.2007 um 23:39 schrieb Alban J Pearce:
>>>>
>>>>> I have installed Angstrom
>>>>> bitbake , nano works
>>>>>
>>>>>  NOTE: build 200701132219: started
>>>>>
>>>>> OE Build Configuration:
>>>>> BB_VERSION     = "1.6.2"
>>>>> OE_REVISION    = "28548b0eda864e1b3a44136bcced77ee8c1e9dab"
>>>>> TARGET_ARCH    = "arm"
>>>>> TARGET_OS      = "linux-gnueabi"
>>>>> MACHINE        = "spitz"
>>>>> DISTRO         = "angstrom"
>>>>> DISTRO_VERSION = "test-20070113"
>>>>> TARGET_FPU     = "soft"
>>>>>
>>>>> But bitbake portsabase fails
>>>>>
>>>>> NOTE: Applying patch 'kernel-keymap.patch'
>>>>> Command Error: exit status: 1  Output:
>>>>> Applying patch kernel-keymap.patch
>>>>> can't find file to patch at input line 10
>>>>> Perhaps you used the wrong -p or --strip option?
>>>>> The text leading up to this was:
>>>>
>>>> Hi,
>>>>
>>>> did you run out of disk space?
> This again highlights why angstrom doesn't support Opie. It is broken.
> Feel free to fix it.
> 
>>
_______________________________________________
Openembedded-devel mailing list
Openembedded-devel@lists.openembedded.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFFqgMWMkyGM64RGpERAvDtAJ4nJODf82sCHHhONrcq/G1BH85/gQCgiUOF
HKgksyYb1LzvXSmz0NTqnvA=
=z+H5
-----END PGP SIGNATURE-----

Re: Angstrom cannot buile portabase

[GoXbox Live]

Well, it seems like there will be a bug once in a while for several
packages, and thats goes also for other gui envirements.
Yesterday i had this bug when trying to build gpe-mini-browser so i guess
there isent just opie thats has some bugs.

http://pastebin.ca/315810

Just get used to it Koen, Opie is still alive and it's still beeing
worked on. The last weeks there have been major updates in the cvs.

gxl


2007/1/14, Koen Kooi <koen@dominion.kabel.utwente.nl>:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> GoXbox Live schreef:
> > Just for information.
> > Opie is not broken. There are now several of us wich build Opie right
> out of
> > oe.dev, without any hacks.
> > So please stop this nonsense about "Opie doesnt work" / "Opie doesnt
> build"
> > and so on.
>
> Yes, I can clearly see how well it works and builds when
> kernel-keymap.patch fails to apply.
>
>
> > Paul Psokolovsky has realy done a great job with Opie in the oe.dev.
> >
> > gxl
> >
> > 2007/1/14, Koen Kooi <koen@dominion.kabel.utwente.nl>:
> > Holger Freyther schreef:
> >>>> Am 13.01.2007 um 23:39 schrieb Alban J Pearce:
> >>>>
> >>>>> I have installed Angstrom
> >>>>> bitbake , nano works
> >>>>>
> >>>>>  NOTE: build 200701132219: started
> >>>>>
> >>>>> OE Build Configuration:
> >>>>> BB_VERSION     = "1.6.2"
> >>>>> OE_REVISION    = "28548b0eda864e1b3a44136bcced77ee8c1e9dab"
> >>>>> TARGET_ARCH    = "arm"
> >>>>> TARGET_OS      = "linux-gnueabi"
> >>>>> MACHINE        = "spitz"
> >>>>> DISTRO         = "angstrom"
> >>>>> DISTRO_VERSION = "test-20070113"
> >>>>> TARGET_FPU     = "soft"
> >>>>>
> >>>>> But bitbake portsabase fails
> >>>>>
> >>>>> NOTE: Applying patch 'kernel-keymap.patch'
> >>>>> Command Error: exit status: 1  Output:
> >>>>> Applying patch kernel-keymap.patch
> >>>>> can't find file to patch at input line 10
> >>>>> Perhaps you used the wrong -p or --strip option?
> >>>>> The text leading up to this was:
> >>>>
> >>>> Hi,
> >>>>
> >>>> did you run out of disk space?
> > This again highlights why angstrom doesn't support Opie. It is broken.
> > Feel free to fix it.
> >
> >>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
> >>
> > _______________________________________________
> > Openembedded-devel mailing list
> > Openembedded-devel@lists.openembedded.org
> > http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.5 (Darwin)
>
> iD8DBQFFqgMWMkyGM64RGpERAvDtAJ4nJODf82sCHHhONrcq/G1BH85/gQCgiUOF
> HKgksyYb1LzvXSmz0NTqnvA=
> =z+H5
> -----END PGP SIGNATURE-----
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: Angstrom cannot buile portabase

[Paul Sokolovsky]

Hello Alban,

Sunday, January 14, 2007, 12:39:31 AM, you wrote:

> I have installed Angstrom 
> bitbake , nano works

[]
> NOTE: Applying patch 'kernel-keymap.patch'
[]
> |+++ qt-2.3.10-snapshot-20060120/src/kernel/qkeyboard_qws.cpp   2006-01-20
> 21:16:25.412761096 +0100
> --------------------------
> No file to patch.  Skipping patch.
> 19 out of 19 hunks ignored
[]

  This patch was done for inconsistent -pN param, I fixed it now, and
checked that it applies and builds.

  At the same time, there're ongoing changes for OPIE in OE.dev, so if
you want to just build apps for your production handheld, OpenZaurus
branch is recommended.

  Otherwise, we'd need Zaurus people to test OPIE changes in OE.dev,
so your help is appreciated, but expect some breakage then. My general
approach to solve current OPIE issues is to reduce machine
discrepancies in its build to zero, set common requirements for a
machine to be supported in OPIE (adhering to which will automatically
mean that machine is supported) and re-review every patch to QT/E for
unavoidability of application in the current situation (that one being
the situation that original OPIE maintainers no longer have
interest/resources to maintain it).

  In this regard, kernel-keymap.patch is in gray area - it currently
being applied inconsistently without even clear major platform
distinction (currently it is applied to Zauruses and Jornada
PocketPC's). So it should be re-evaluated for benefits it offer vs
possible requirements it has and maintenance burden, and either being
applied consistently for all machines, or dropped.

  I wanted to discuss these issues with Holger (zecke) (and other core
OPIE folks who'd have time/interest) on IRC first, but now that this
issue was raised, I dump this "seed" message here...



-- 
Best regards,
 Paul                            mailto:pmiscml@gmail.com

Re: Gtk+ 2.6.10 w/ newer Glibs

[Michael 'Mickey' Lauer]

Koen Kooi wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1

> Richard Purdie schreef:

>> To fix this there needs to be a patch that removes
>> -DG_DISABLE_DEPRECATED from every Makefile.am.

> patch? sed! :)

Did that and it seems to work -- thanks!

Regards,

:M:
-- 
Michael 'Mickey' Lauer | IT-Freelancer | http://www.vanille-media.de