The rpurdie/work-in-progress branch has been merged

The rpurdie/work-in-progress branch has been merged

[Richard Purdie]

I finally felt the time was right so I've merged the work-in-progress
branch I've been talking about to OE.dev. This contains various
architecture improvements and legacy staging cleanups that have been
used in Poky for a while with great benefit.

One thing people have seen with this branch is configure failures like:


"""
Written by Tom Tromey <tromey@redhat.com>
       and Alexandre Duret-Lutz <adl@gnu.org>.
AUTOV is 1.10
cp: cannot stat `/media/data1/builds/oe1/tmp/sysroots/armv5te-oe-linux-gnueabi/usr/share/gettext/config.rpath': No such file or directory
"""

This means you need to inherit the gettext class in the .bb file as the
package has a gettext dependency that is missing, i.e. a trivial fix.

Obviously these don't appear once gettext has been built as a dependency
for any one thing. There are warnings about missing inherits in the QA
tests though. Poky has these as errors, I don't think OE is quite ready
for that yet but it would be something to start working on.

I and others will keep an eye open for problems related to this merge
and address them as quickly as possible. 

These changes are important to move us closer to our goals of improved
sysroot handling, better packaged staging and so forth.

Cheers,

Richard

Re: The rpurdie/work-in-progress branch has been merged

[Tom Rini]

On Thu, 2010-05-06 at 00:43 +0100, Richard Purdie wrote:
> I finally felt the time was right so I've merged the work-in-progress
> branch I've been talking about to OE.dev. This contains various
> architecture improvements and legacy staging cleanups that have been
> used in Poky for a while with great benefit.

I've started testing DISTRO=minimal with the machines and images I said
earlier.  Will report eventually (it's a lotta stuff).

-- 
Tom Rini <tom_rini@mentor.com>
Mentor Graphics Corporation

Re: The rpurdie/work-in-progress branch has been merged

[Khem Raj]

On Wed, May 5, 2010 at 5:14 PM, Tom Rini <tom_rini@mentor.com> wrote:
> On Thu, 2010-05-06 at 00:43 +0100, Richard Purdie wrote:
>> I finally felt the time was right so I've merged the work-in-progress
>> branch I've been talking about to OE.dev. This contains various
>> architecture improvements and legacy staging cleanups that have been
>> used in Poky for a while with great benefit.
>
> I've started testing DISTRO=minimal with the machines and images I said
> earlier.  Will report eventually (it's a lotta stuff).

I will testout various arches with uclibc
>
> --
> Tom Rini <tom_rini@mentor.com>
> Mentor Graphics Corporation
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: The rpurdie/work-in-progress branch has been merged

[Koen Kooi]

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

On 06-05-10 01:43, Richard Purdie wrote:

> Obviously these don't appear once gettext has been built as a dependency
> for any one thing. There are warnings about missing inherits in the QA
> tests though. Poky has these as errors, I don't think OE is quite ready
> for that yet but it would be something to start working on.

You do know that only angstrom cares about QA, right? It's the only
distro in OE that is using insane.bbclass...
Maybe the TSC can make that class mandatory :)

regards,

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

iD8DBQFL4l2HMkyGM64RGpERAq9TAKC5T7006+TYFcDJ1K+FmsdaxvHVAACfUhxy
7vIvsCN1wqzG+ner/9tZmi0=
=PlqM
-----END PGP SIGNATURE-----

Re: The rpurdie/work-in-progress branch has been merged

[Koen Kooi]

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

On 06-05-10 01:43, Richard Purdie wrote:
> I finally felt the time was right so I've merged the work-in-progress
> branch I've been talking about to OE.dev. This contains various
> architecture improvements and legacy staging cleanups that have been
> used in Poky for a while with great benefit.

With bitbake 1.10 I get:

NOTE: Preparing runqueue
Traceback (most recent call last):
  File "/OE/bitbake/lib/bb/server/none.py", line 128, in idle_commands
    retval = function(self, data, False)
  File "/OE/bitbake/lib/bb/cooker.py", line 730, in buildTargetsIdle
    retval = rq.execute_runqueue()
  File "/OE/bitbake/lib/bb/runqueue.py", line 830, in execute_runqueue
    self.prepare_runqueue()
  File "/OE/bitbake/lib/bb/runqueue.py", line 513, in prepare_runqueue
    if len(runq_recrdepends[task]) > 0:
IndexError: list index out of range

when doing MACHINE=beagleboard bitbake console-image

regards,

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

iD8DBQFL4mBcMkyGM64RGpERAgqjAJ0aTRFgft4TINZP181N4jBgfOb2wQCfQMZD
2JHtNB/w4uDZJBkJMCcfyic=
=60rt
-----END PGP SIGNATURE-----

Re: The rpurdie/work-in-progress branch has been merged

[Koen Kooi]

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

On 06-05-10 01:43, Richard Purdie wrote:
> I finally felt the time was right so I've merged the work-in-progress
> branch I've been talking about to OE.dev. This contains various
> architecture improvements and legacy staging cleanups that have been
> used in Poky for a while with great benefit.

Khem, can you fix the gcc 4.5.0 recipes, Richard didn't convert those to
the new style gcc stuff. It should a matter of shuffling includes around.

regards,

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

iD8DBQFL4mPgMkyGM64RGpERAr/PAJ9dN6l44gHs7cK6PApqxt32Iq2fkACgjEb4
wqrN1+xlKp8Ps2o+lOdWfl8=
=pMUA
-----END PGP SIGNATURE-----

Re: The rpurdie/work-in-progress branch has been merged

[Koen Kooi]

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

On 06-05-10 08:23, Koen Kooi wrote:
> On 06-05-10 01:43, Richard Purdie wrote:
>> I finally felt the time was right so I've merged the work-in-progress
>> branch I've been talking about to OE.dev. This contains various
>> architecture improvements and legacy staging cleanups that have been
>> used in Poky for a while with great benefit.
> 
> With bitbake 1.10 I get:
> 
> NOTE: Preparing runqueue
> Traceback (most recent call last):
>   File "/OE/bitbake/lib/bb/server/none.py", line 128, in idle_commands
>     retval = function(self, data, False)
>   File "/OE/bitbake/lib/bb/cooker.py", line 730, in buildTargetsIdle
>     retval = rq.execute_runqueue()
>   File "/OE/bitbake/lib/bb/runqueue.py", line 830, in execute_runqueue
>     self.prepare_runqueue()
>   File "/OE/bitbake/lib/bb/runqueue.py", line 513, in prepare_runqueue
>     if len(runq_recrdepends[task]) > 0:
> IndexError: list index out of range
> 
> when doing MACHINE=beagleboard bitbake console-image

Bitbake 1.8, 1.10 and master exhibit the same behaviour. 'bitbake gcc'
does work, trying to narrow down which recipe is causing this, it's not
the toolchain.

regards,

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

iD8DBQFL4mmqMkyGM64RGpERAuWNAJ4k+PBIWwAZwb9rhnIL3xX7AHwYAwCgt0A2
+Li/nkosj5HWBLeAideTYnQ=
=HcC1
-----END PGP SIGNATURE-----

Re: The rpurdie/work-in-progress branch has been merged

[Khem Raj]

On Wed, May 5, 2010 at 11:38 PM, Koen Kooi <k.kooi@student.utwente.nl> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 06-05-10 01:43, Richard Purdie wrote:
>> I finally felt the time was right so I've merged the work-in-progress
>> branch I've been talking about to OE.dev. This contains various
>> architecture improvements and legacy staging cleanups that have been
>> used in Poky for a while with great benefit.
>
> Khem, can you fix the gcc 4.5.0 recipes, Richard didn't convert those to
> the new style gcc stuff. It should a matter of shuffling includes around.
>

yes and I will also have to do eglibc.

> regards,
>
> Koen
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.5 (Darwin)
>
> iD8DBQFL4mPgMkyGM64RGpERAr/PAJ9dN6l44gHs7cK6PApqxt32Iq2fkACgjEb4
> wqrN1+xlKp8Ps2o+lOdWfl8=
> =pMUA
> -----END PGP SIGNATURE-----
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: The rpurdie/work-in-progress branch has been merged

[Richard Purdie]

On Thu, 2010-05-06 at 09:03 +0200, Koen Kooi wrote:
> On 06-05-10 08:23, Koen Kooi wrote:
> > On 06-05-10 01:43, Richard Purdie wrote:
> >> I finally felt the time was right so I've merged the work-in-progress
> >> branch I've been talking about to OE.dev. This contains various
> >> architecture improvements and legacy staging cleanups that have been
> >> used in Poky for a while with great benefit.
> > 
> > With bitbake 1.10 I get:
> > 
> > NOTE: Preparing runqueue
> > Traceback (most recent call last):
> >   File "/OE/bitbake/lib/bb/server/none.py", line 128, in idle_commands
> >     retval = function(self, data, False)
> >   File "/OE/bitbake/lib/bb/cooker.py", line 730, in buildTargetsIdle
> >     retval = rq.execute_runqueue()
> >   File "/OE/bitbake/lib/bb/runqueue.py", line 830, in execute_runqueue
> >     self.prepare_runqueue()
> >   File "/OE/bitbake/lib/bb/runqueue.py", line 513, in prepare_runqueue
> >     if len(runq_recrdepends[task]) > 0:
> > IndexError: list index out of range
> > 
> > when doing MACHINE=beagleboard bitbake console-image
> 
> Bitbake 1.8, 1.10 and master exhibit the same behaviour. 'bitbake gcc'
> does work, trying to narrow down which recipe is causing this, it's not
> the toolchain.

Someone just told me OE was broken as this email hadn't been replied
too. For reference, Koen and I talked elsewhere, a fix was pushed and
this issue was resolved. 

To further update, Khem pushed updates for eglibc and gcc 4.5 (thanks).
Koen also pushed some legacy staging removals which had some issues but
these have also all been fixed. There was also an error appearing about
"bb" being unknown and that is also fixed.

So in summary, we don't have any known problems due to these changes at
the moment :).

I wasn't clear in my original email but all issues that people raised
testing the branch had been addressed before it was pushed. Thanks go to
the people who did test it.

Cheers,

Richard

Re: The rpurdie/work-in-progress branch has been merged

[Koen Kooi]

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

On 06-05-10 14:10, Richard Purdie wrote:
> On Thu, 2010-05-06 at 09:03 +0200, Koen Kooi wrote:
>> On 06-05-10 08:23, Koen Kooi wrote:
>>> On 06-05-10 01:43, Richard Purdie wrote:
>>>> I finally felt the time was right so I've merged the work-in-progress
>>>> branch I've been talking about to OE.dev. This contains various
>>>> architecture improvements and legacy staging cleanups that have been
>>>> used in Poky for a while with great benefit.
>>>
>>> With bitbake 1.10 I get:
>>>
>>> NOTE: Preparing runqueue
>>> Traceback (most recent call last):
>>>   File "/OE/bitbake/lib/bb/server/none.py", line 128, in idle_commands
>>>     retval = function(self, data, False)
>>>   File "/OE/bitbake/lib/bb/cooker.py", line 730, in buildTargetsIdle
>>>     retval = rq.execute_runqueue()
>>>   File "/OE/bitbake/lib/bb/runqueue.py", line 830, in execute_runqueue
>>>     self.prepare_runqueue()
>>>   File "/OE/bitbake/lib/bb/runqueue.py", line 513, in prepare_runqueue
>>>     if len(runq_recrdepends[task]) > 0:
>>> IndexError: list index out of range
>>>
>>> when doing MACHINE=beagleboard bitbake console-image
>>
>> Bitbake 1.8, 1.10 and master exhibit the same behaviour. 'bitbake gcc'
>> does work, trying to narrow down which recipe is causing this, it's not
>> the toolchain.
> 
> Someone just told me OE was broken as this email hadn't been replied
> too. For reference, Koen and I talked elsewhere, a fix was pushed and
> this issue was resolved. 

I have built console-image for hawkboard, smartq5 and beagleboard and
they all work as well as before. Thanks to the removal of legacy staging
in gcc, glibc, etc, the build was a bit faster (but still dominated by
localegen).

> To further update, Khem pushed updates for eglibc and gcc 4.5 (thanks).
> Koen also pushed some legacy staging removals which had some issues but
> these have also all been fixed.

Note to self: if something doesn't use 'make install', it needs
'NATIVE_INSTALL_WORKS = 1' when using BBCLASSEXTEND = native.

regards,

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

iD8DBQFL4rqFMkyGM64RGpERAotJAJ4+LvJTHsPCfhRPcYCmHRYbzgo/5gCeKBvz
VKoBx4YqEtsQiONmXlDxBQ8=
=5V/p
-----END PGP SIGNATURE-----

Re: The rpurdie/work-in-progress branch has been merged

[Martin Jansa]

On Thu, May 06, 2010 at 01:10:27PM +0100, Richard Purdie wrote:
> Someone just told me OE was broken as this email hadn't been replied
> too. For reference, Koen and I talked elsewhere, a fix was pushed and
> this issue was resolved. 

No OE is great and we know it :).

> To further update, Khem pushed updates for eglibc and gcc 4.5 (thanks).
> Koen also pushed some legacy staging removals which had some issues but
> these have also all been fixed. There was also an error appearing about
> "bb" being unknown and that is also fixed.

I've rebuilt 2 images today and except those smaller issues as you said 
and which were fixed in minutes everything went OK.

> So in summary, we don't have any known problems due to these changes at
> the moment :).
>
> I wasn't clear in my original email but all issues that people raised
> testing the branch had been addressed before it was pushed. Thanks go to
> the people who did test it.

I haven't test it before push and from that announcement I expected some
problems, but the way it was done was great, well prepared, work of
professionals, thanks to all people involved.

Cheers,

-- 
uin:136542059                jid:Martin.Jansa@gmail.com
Jansa Martin                 sip:jamasip@voip.wengo.fr 
JaMa                         

Re: The rpurdie/work-in-progress branch has been merged

[Tom Rini]

On Wed, 2010-05-05 at 17:14 -0700, Tom Rini wrote:
> On Thu, 2010-05-06 at 00:43 +0100, Richard Purdie wrote:
> > I finally felt the time was right so I've merged the work-in-progress
> > branch I've been talking about to OE.dev. This contains various
> > architecture improvements and legacy staging cleanups that have been
> > used in Poky for a while with great benefit.
> 
> I've started testing DISTRO=minimal with the machines and images I said
> earlier.  Will report eventually (it's a lotta stuff).

For DISTRO=minimal, mpc8315e-rdb and calamari fail, both before and
after these changes.  For qemux86, qemuarm, qemumips and db1200,
minimal-image builds and for the QEMU boards, x11-image builds as well.

-- 
Tom Rini <tom_rini@mentor.com>
Mentor Graphics Corporation

Re: The rpurdie/work-in-progress branch has been merged

[Khem Raj]

On Thu, May 6, 2010 at 12:52 PM, Tom Rini <tom_rini@mentor.com> wrote:
> On Wed, 2010-05-05 at 17:14 -0700, Tom Rini wrote:
>> On Thu, 2010-05-06 at 00:43 +0100, Richard Purdie wrote:
>> > I finally felt the time was right so I've merged the work-in-progress
>> > branch I've been talking about to OE.dev. This contains various
>> > architecture improvements and legacy staging cleanups that have been
>> > used in Poky for a while with great benefit.
>>
>> I've started testing DISTRO=minimal with the machines and images I said
>> earlier.  Will report eventually (it's a lotta stuff).
>
> For DISTRO=minimal, mpc8315e-rdb and calamari fail, both before and
> after these changes.

is it the same error about missing function from libgcc or something else ?


  For qemux86, qemuarm, qemumips and db1200,
> minimal-image builds and for the QEMU boards, x11-image builds as well.
>
> --
> Tom Rini <tom_rini@mentor.com>
> Mentor Graphics Corporation
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: The rpurdie/work-in-progress branch has been merged

[Michael Lippautz]

FYI: Console-/minimal/base-Image built fine for IGEPv2 board (build
from scratch)

Packaged-staging: Destination of packages is now defined by PSTAGE_DIR.

Regards,
Michael

Re: The rpurdie/work-in-progress branch has been merged

[Tom Rini]

On Thu, 2010-05-06 at 18:17 -0700, Khem Raj wrote:
> On Thu, May 6, 2010 at 12:52 PM, Tom Rini <tom_rini@mentor.com> wrote:
> > On Wed, 2010-05-05 at 17:14 -0700, Tom Rini wrote:
> >> On Thu, 2010-05-06 at 00:43 +0100, Richard Purdie wrote:
> >> > I finally felt the time was right so I've merged the work-in-progress
> >> > branch I've been talking about to OE.dev. This contains various
> >> > architecture improvements and legacy staging cleanups that have been
> >> > used in Poky for a while with great benefit.
> >>
> >> I've started testing DISTRO=minimal with the machines and images I said
> >> earlier.  Will report eventually (it's a lotta stuff).
> >
> > For DISTRO=minimal, mpc8315e-rdb and calamari fail, both before and
> > after these changes.
> 
> is it the same error about missing function from libgcc or something else ?

I'll post some new threads with pastebins of what it looks likes.

-- 
Tom Rini <tom_rini@mentor.com>
Mentor Graphics Corporation

Re: The rpurdie/work-in-progress branch has been merged

[Andrea Adami]

If someone can find it useful:

andrea@mizar /oe $ grep -n 'inherit gettext' oelog -B1
494-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/i686-linux/coreutils-native-7.2-r1/coreutils-7.2/configure.ac.
495:Missing inherit gettext?
--
5755-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/gcc-cross-initial-4.3.3-r11.1/gcc-4.3.3/intl/configure.ac.
5756:Missing inherit gettext?
--
6202-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/gcc-cross-intermediate-4.3.3-r11.1/gcc-4.3.3/intl/configure.ac.
6203:Missing inherit gettext?
--
9241-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/gcc-cross-4.3.3-r11.1/gcc-4.3.3/intl/configure.ac.
9242:Missing inherit gettext?
--
23576-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/gstreamer-0.10.28-r0/gstreamer-0.10.28/configure.ac.
23577:Missing inherit gettext?
--
25826-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/gst-plugins-base-0.10.28-r8.1/gst-plugins-base-0.10.28/configure.ac.
25827:Missing inherit gettext?
25828-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/gst-plugins-base-0.10.28-r8.1/gst-plugins-base-0.10.28/.pc/gst-plugins-base_rowstride.patch/configure.ac.
25829:Missing inherit gettext?
--
34842-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/alsa-utils-1.0.20-r1/alsa-utils-1.0.20/configure.in.
34843:Missing inherit gettext?
--
35009-NOTE: Gettext required but not in DEPENDS for file
/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/e2fsprogs-1.41.9-r23/e2fsprogs-1.41.9/configure.in.
35010:Missing inherit gettext?
andrea@mizar /oe $

Re: The rpurdie/work-in-progress branch has been merged

[Andrea Adami]

another anomaly, after rebuild from scratch, appears already on task 17:

NOTE: Running task 17 of 5034 (ID: 1135,
/oe/openembedded/recipes/stage-manager/stagemanager-native_0.0.1.bb,
do_install)
cp: cannot stat
`/oe/build/tmp/work/i686-linux/stagemanager-native-0.0.1-r12/sysroot-destdir//oe/build/tmp/*':
No such file or directory

Andrea

Re: The rpurdie/work-in-progress branch has been merged

[Andrea Adami]

Sorry, these are all occurrences (console image + opie image
+x11-image). Bad do_install ?

69:cp: cannot stat
`/oe/build/tmp/work/i686-linux/stagemanager-native-0.0.1-r12/sysroot-destdir//oe/build/tmp/*':
No such file or directory
778:cp: cannot stat
`/oe/build/tmp/work/i686-linux/quilt-native-0.48-r0/sysroot-destdir//oe/build/tmp/*':
No such file or directory
6276:cp: cannot stat
`/oe/build/tmp/work/i686-linux/cmake-native-2.6.4-r2/sysroot-destdir//oe/build/tmp/*':
No such file or directory
6649:cp: cannot stat
`/oe/build/tmp/work/i686-linux/tcl-native-8.4.19-r4/image/*': No such
file or directory
28082:cp: cannot stat
`/oe/build/tmp/work/c7x0-angstrom-linux-gnueabi/task-boot-1.0-r52/image/*':
No such file or directory
29410:cp: cannot stat
`/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/opie-libqrsync-1.2.4-r0/image/*':
No such file or directory
33418:cp: cannot stat
`/oe/build/tmp/work/c7x0-angstrom-linux-gnueabi/task-opie-1.0-r13/image/*':
No such file or directory
33468:cp: cannot stat
`/oe/build/tmp/work/c7x0-angstrom-linux-gnueabi/task-base-1.0-r89/image/*':
No such file or directory
35204:cp: cannot stat
`/oe/build/tmp/work/all-angstrom-linux-gnueabi/angstrom-libc-fixup-hack-1.0-r0/image/*':
No such file or directory
38173:cp: cannot stat
`/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/xscrnsaverh-1.0-r0/image/*':
No such file or directory
39753:cp: cannot stat
`/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/matchbox-0.9.1-r1/image/*':
No such file or directory
43115:cp: cannot stat
`/oe/build/tmp/work/all-angstrom-linux-gnueabi/angstrom-gpe-task-settings-1.0-r35/image/*':
No such file or directory
43223:cp: cannot stat
`/oe/build/tmp/work/c7x0-angstrom-linux-gnueabi/angstrom-gpe-task-base-1.0-r40/image/*':
No such file or directory
43315:cp: cannot stat
`/oe/build/tmp/work/c7x0-angstrom-linux-gnueabi/angstrom-x11-base-depends-1.0-r43/image/*':
No such file or directory


Andrea