testing branch 2010-10-08

testing branch 2010-10-08

[Cliff Brake]

master branch has been merged to testing-next and is ready for clean builds.

Last week, we ran into quite a few problems, so I'm not tagging or
moving the testing branch forward.

Note, we have moved the testing cycle from Fri to Fri instead of Mon
to Mon.  This is because people more likely have machines with spare
cycles over the weekend.

http://wiki.openembedded.net/index.php/Testing

Thanks,
Cliff

-- 
=================
http://bec-systems.com

Re: testing branch 2010-10-08

[Frans Meulenbroeks]

2010/10/8 Cliff Brake <cliff.brake@gmail.com>:
> master branch has been merged to testing-next and is ready for clean builds.
>
> Last week, we ran into quite a few problems, so I'm not tagging or
> moving the testing branch forward.
>
> Note, we have moved the testing cycle from Fri to Fri instead of Mon
> to Mon.  This is because people more likely have machines with spare
> cycles over the weekend.
>
> http://wiki.openembedded.net/index.php/Testing
>
> Thanks,
> Cliff
>

Thanks for doing this Cliff.
I've got my 6 builds already scheduled on our autobuilder.I will
report back on the wiki.

One small remark: timewise, I would prefer creation of the branch of
your friday morning. For us europeans the weekend is already started.
Doing it a little bit earlier allows people like Steffen and me to
kick it off before we go home.
Alternately it might help if we can define a time before which the
branch is made. That would allow a cron jub or so to kick things off.

Frans

Re: testing branch 2010-10-08

[Frans Meulenbroeks]

libgee-native configure is failing:
http://tinderbox.openembedded.net/public/logs/task/8568667.txt

Re: testing branch 2010-10-08

[Michael 'Mickey' Lauer]

Am Samstag, den 09.10.2010, 22:58 +0200 schrieb Frans Meulenbroeks:
> libgee-native configure is failing:
> http://tinderbox.openembedded.net/public/logs/task/8568667.txt

Does it contain 72c2ac940065948cb688fde821e51efac6b71b4b, which was
supposed to fix that?

Cheers,
-- 
:M:

Re: testing branch 2010-10-08

[Graham Gower]

On 9 October 2010 17:40, Frans Meulenbroeks <fransmeulenbroeks@gmail.com> wrote:
> Thanks for doing this Cliff.
> I've got my 6 builds already scheduled on our autobuilder.I will
> report back on the wiki.
>
> One small remark: timewise, I would prefer creation of the branch of
> your friday morning. For us europeans the weekend is already started.
> Doing it a little bit earlier allows people like Steffen and me to
> kick it off before we go home.

How about Thursday, that way us UTC+9:30ians (+1hour daylight
savings!) can kick it off before our weekend :)

> Alternately it might help if we can define a time before which the
> branch is made. That would allow a cron jub or so to kick things off.

I like this idea too.

-Graham

Re: testing branch 2010-10-08

[Paul Menzel]

[-- Attachment #1: Type: text/plain, Size: 623 bytes --]

Am Sonntag, den 10.10.2010, 00:55 +0200 schrieb Michael 'Mickey' Lauer:
> Am Samstag, den 09.10.2010, 22:58 +0200 schrieb Frans Meulenbroeks:
> > libgee-native configure is failing:
> > http://tinderbox.openembedded.net/public/logs/task/8568667.txt
> 
> Does it contain 72c2ac940065948cb688fde821e51efac6b71b4b, which was
> supposed to fix that?

That is the exact same issues I reported 12 hours earlier [1].


Thanks,

Paul


[1] http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-October/025222.html
    "DEPENDS not honored: `do_configure` runs before `do_populate_sysroot` of dependencies"

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: testing branch 2010-10-08

[Khem Raj]

On Sun, Oct 10, 2010 at 1:20 AM, Paul Menzel
<paulepanter@users.sourceforge.net> wrote:
> Am Sonntag, den 10.10.2010, 00:55 +0200 schrieb Michael 'Mickey' Lauer:
>> Am Samstag, den 09.10.2010, 22:58 +0200 schrieb Frans Meulenbroeks:
>> > libgee-native configure is failing:
>> > http://tinderbox.openembedded.net/public/logs/task/8568667.txt
>>
>> Does it contain 72c2ac940065948cb688fde821e51efac6b71b4b, which was
>> supposed to fix that?
>
> That is the exact same issues I reported 12 hours earlier [1].
>
>

I fixed it by not reautoconfing for now I think it should be
good enough.

> Thanks,
>
> Paul
>
>
> [1] http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-October/025222.html
>    "DEPENDS not honored: `do_configure` runs before `do_populate_sysroot` of dependencies"
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>
>

Re: testing branch 2010-10-08

[Paul Menzel]

[-- Attachment #1: Type: text/plain, Size: 1584 bytes --]

Am Sonntag, den 10.10.2010, 02:13 -0700 schrieb Khem Raj:
> On Sun, Oct 10, 2010 at 1:20 AM, Paul Menzel wrote:
> > Am Sonntag, den 10.10.2010, 00:55 +0200 schrieb Michael 'Mickey' Lauer:
> >> Am Samstag, den 09.10.2010, 22:58 +0200 schrieb Frans Meulenbroeks:
> >> > libgee-native configure is failing:
> >> > http://tinderbox.openembedded.net/public/logs/task/8568667.txt
> >>
> >> Does it contain 72c2ac940065948cb688fde821e51efac6b71b4b, which was
> >> supposed to fix that?
> >
> > That is the exact same issues I reported 12 hours earlier [1].
> 
> I fixed it by not reautoconfing for now I think it should be
> good enough.

Thank you for doing that. This is commit 88cb82 [2]. 

        libgee_0.5.2.bb/libgee_0.6.0.bb: Fix native recipe build

Could you please at least add a link to the list thread describing the
problem to the commit message next time, so people know what is being
fixed. I know it takes more time and I should be glad that you are so
responsive and that you fix things quickly. And I am thankful. But in
the end it will save other people’s time if those things are documented
properly.

And still, the other thread [1] should find a conclusion, since for me
it looks like the current behavior is wrong.


Thanks,

Paul


> > [1] http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-October/025222.html
> >    "DEPENDS not honored: `do_configure` runs before `do_populate_sysroot` of dependencies"

[2] http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=88cb8237bcad77fc2a64a1e84b6be63c94ce8b58

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: testing branch 2010-10-08

[Khem Raj]

On Sun, Oct 10, 2010 at 2:38 AM, Paul Menzel
<paulepanter@users.sourceforge.net> wrote:
> Am Sonntag, den 10.10.2010, 02:13 -0700 schrieb Khem Raj:
>> On Sun, Oct 10, 2010 at 1:20 AM, Paul Menzel wrote:
>> > Am Sonntag, den 10.10.2010, 00:55 +0200 schrieb Michael 'Mickey' Lauer:
>> >> Am Samstag, den 09.10.2010, 22:58 +0200 schrieb Frans Meulenbroeks:
>> >> > libgee-native configure is failing:
>> >> > http://tinderbox.openembedded.net/public/logs/task/8568667.txt
>> >>
>> >> Does it contain 72c2ac940065948cb688fde821e51efac6b71b4b, which was
>> >> supposed to fix that?
>> >
>> > That is the exact same issues I reported 12 hours earlier [1].
>>
>> I fixed it by not reautoconfing for now I think it should be
>> good enough.
>
> Thank you for doing that. This is commit 88cb82 [2].
>
>        libgee_0.5.2.bb/libgee_0.6.0.bb: Fix native recipe build
>
> Could you please at least add a link to the list thread describing the
> problem to the commit message next time, so people know what is being
> fixed.

while the details are in the mail and I notified the mail thread soon after
the fix was pushed isnt that enough of a notice for a trivial problem.

 I know it takes more time and I should be glad that you are so
> responsive and that you fix things quickly. And I am thankful. But in
> the end it will save other people’s time if those things are documented
> properly.
>
> And still, the other thread [1] should find a conclusion, since for me
> it looks like the current behavior is wrong.

what behavior is wrong ?

Re: testing branch 2010-10-08

[Paul Menzel]

[-- Attachment #1: Type: text/plain, Size: 423 bytes --]

Am Sonntag, den 10.10.2010, 12:29 -0700 schrieb Khem Raj:
> On Sun, Oct 10, 2010 at 2:38 AM, Paul Menzel wrote:

[…]

> > And still, the other thread [1] should find a conclusion, since for me
> > it looks like the current behavior is wrong.
> 
> what behavior is wrong ?

As the thread subject says: DEPENDS not honored: `do_configure` runs
before `do_populate_sysroot` of dependencies.


Thanks,

Paul

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

commit messages (was: testing branch 2010-10-08)

[Paul Menzel]

[-- Attachment #1: Type: text/plain, Size: 2020 bytes --]

Am Sonntag, den 10.10.2010, 12:29 -0700 schrieb Khem Raj:
> On Sun, Oct 10, 2010 at 2:38 AM, Paul Menzel wrote:
> > Am Sonntag, den 10.10.2010, 02:13 -0700 schrieb Khem Raj:
> >> On Sun, Oct 10, 2010 at 1:20 AM, Paul Menzel wrote:
> >> > Am Sonntag, den 10.10.2010, 00:55 +0200 schrieb Michael 'Mickey' Lauer:
> >> >> Am Samstag, den 09.10.2010, 22:58 +0200 schrieb Frans Meulenbroeks:
> >> >> > libgee-native configure is failing:
> >> >> > http://tinderbox.openembedded.net/public/logs/task/8568667.txt
> >> >>
> >> >> Does it contain 72c2ac940065948cb688fde821e51efac6b71b4b, which was
> >> >> supposed to fix that?
> >> >
> >> > That is the exact same issues I reported 12 hours earlier [1].
> >>
> >> I fixed it by not reautoconfing for now I think it should be
> >> good enough.
> >
> > Thank you for doing that. This is commit 88cb82 [2].
> >
> >        libgee_0.5.2.bb/libgee_0.6.0.bb: Fix native recipe build
> >
> > Could you please at least add a link to the list thread describing the
> > problem to the commit message next time, so people know what is being
> > fixed.
> 
> while the details are in the mail and I notified the mail thread soon after
> the fix was pushed isnt that enough of a notice for a trivial problem.

I was more looking from the perspective of people going through the
commit history and not the people reading the list.

I came across several cases where the commit message says to fix a
build, but there was no indication what the error was in the first
place. It is even more important when going through the changes after
time.

So I would be grateful if everyone could adhere to the commit policy [1]
as documentation is in my opinion one of the most important things when
more people are involved.

I would even require that everyone has to build test a change at least
once in a clean environment and document what distribution and what
machine it was built for.

[…]


[1] http://wiki.openembedded.net/index.php/Commit_Policy

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: testing branch 2010-10-08

[Khem Raj]

On Sun, Oct 10, 2010 at 12:37 PM, Paul Menzel
<paulepanter@users.sourceforge.net> wrote:
> Am Sonntag, den 10.10.2010, 12:29 -0700 schrieb Khem Raj:
>> On Sun, Oct 10, 2010 at 2:38 AM, Paul Menzel wrote:
>
> […]
>
>> > And still, the other thread [1] should find a conclusion, since for me
>> > it looks like the current behavior is wrong.
>>
>> what behavior is wrong ?
>
> As the thread subject says: DEPENDS not honored: `do_configure` runs
> before `do_populate_sysroot` of dependencies.

hmmm I doubt it. We should see such problems more often if this was
happening. It might be a special case in what you are seeing. You
should
look through task dependency tree and try to see if some task is not
defined correctly.

>
>
> Thanks,
>
> Paul
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>
>

Re: testing branch 2010-10-08

[Sledz, Steffen]

do_configure still fails for epiphany-2.30.2-r3 (see
<http://tinderbox.openembedded.net/packages/835265/>) with:

"checking location of system Certificate Authority list...
configure: error: could not find. Use --with-ca-file=path
to set, or --without-ca-file to disable"

I thought the additional ca-certifactes dependency (see thread
<http://thread.gmane.org/gmane.comp.handhelds.openembedded/37453/focus=37585>)
fixes this, but it does not seem so. :(

I'd a look into my openSUSE-system and there epiphany uses
a file /var/lib/ca-certificates/ca-bundle.pem (linked to
/etc/ssl/ca-bundle.pem) installed by the ca-certificates package.

The OE ca-certificates package does not install such a bundle.
So may be that's the real problem.

Steffen

Re: testing branch 2010-10-08

[Sledz, Steffen]

On 10/11/2010 09:34 AM, Sledz, Steffen wrote:
> do_configure still fails for epiphany-2.30.2-r3 (see
> <http://tinderbox.openembedded.net/packages/835265/>) with:
>
> "checking location of system Certificate Authority list...
> configure: error: could not find. Use --with-ca-file=path
> to set, or --without-ca-file to disable"
>
> I thought the additional ca-certifactes dependency (see thread
> <http://thread.gmane.org/gmane.comp.handhelds.openembedded/37453/focus=37585>)
> fixes this, but it does not seem so. :(
>
> I'd a look into my openSUSE-system and there epiphany uses
> a file /var/lib/ca-certificates/ca-bundle.pem (linked to
> /etc/ssl/ca-bundle.pem) installed by the ca-certificates package.
>
> The OE ca-certificates package does not install such a bundle.
> So may be that's the real problem.

A look into the ca-certificates recipe shows this:

  pkg_postinst_${PN} () {
          /usr/sbin/update-ca-certificates
  }

I think this is to late if another package needs the result from this call in its configuration stage.

Steffen

Re: testing branch 2010-10-08

[Koen Kooi]

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

On 11-10-10 17:34, Sledz, Steffen wrote:
> On 10/11/2010 09:34 AM, Sledz, Steffen wrote:
>> do_configure still fails for epiphany-2.30.2-r3 (see
>> <http://tinderbox.openembedded.net/packages/835265/>) with:
>>
>> "checking location of system Certificate Authority list...
>> configure: error: could not find. Use --with-ca-file=path
>> to set, or --without-ca-file to disable"
>>
>> I thought the additional ca-certifactes dependency (see thread
>> <http://thread.gmane.org/gmane.comp.handhelds.openembedded/37453/focus=37585>)
>> fixes this, but it does not seem so. :(
>>
>> I'd a look into my openSUSE-system and there epiphany uses
>> a file /var/lib/ca-certificates/ca-bundle.pem (linked to
>> /etc/ssl/ca-bundle.pem) installed by the ca-certificates package.
>>
>> The OE ca-certificates package does not install such a bundle.
>> So may be that's the real problem.
> 
> A look into the ca-certificates recipe shows this:
> 
>   pkg_postinst_${PN} () {
>           /usr/sbin/update-ca-certificates
>   }
> 
> I think this is to late if another package needs the result from this call in its configuration stage.

That's missing a $D check as well, since in that form it can't be run
off-line

regards,

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

iD8DBQFMszX9MkyGM64RGpERAntxAJ4ks0QTGm4TYe0I4oFZkJkVibnYxQCfX383
lMq5v0la+u26N9XyIUdTVFk=
=6Bnt
-----END PGP SIGNATURE-----

Re: commit messages (was: testing branch 2010-10-08)

[Cliff Brake]

On Sun, Oct 10, 2010 at 3:50 PM, Paul Menzel
<paulepanter@users.sourceforge.net> wrote:
> I would even require that everyone has to build test a change at least
> once in a clean environment and document what distribution and what
> machine it was built for.

Concerning the suggestion that "changes must be built once in a clean
environment," I'm hoping the testing effort will help catch things
like this.  My thought is not to put too much burden on people making
constructive changes, but yet have a way that problems get found in a
reasonable time frame.

Cliff

-- 
=================
http://bec-systems.com

Re: testing branch 2010-10-08

[Sledz, Steffen]

Am 11.10.2010 18:06, schrieb Koen Kooi:
> On 11-10-10 17:34, Sledz, Steffen wrote:
>> On 10/11/2010 09:34 AM, Sledz, Steffen wrote:
>>> do_configure still fails for epiphany-2.30.2-r3 (see
>>> <http://tinderbox.openembedded.net/packages/835265/>) with:
>>>
>>> "checking location of system Certificate Authority list...
>>> configure: error: could not find. Use --with-ca-file=path
>>> to set, or --without-ca-file to disable"
>>>
>>> I thought the additional ca-certifactes dependency (see thread
>>> <http://thread.gmane.org/gmane.comp.handhelds.openembedded/37453/focus=37585>)
>>> fixes this, but it does not seem so. :(
>>>
>>> I'd a look into my openSUSE-system and there epiphany uses
>>> a file /var/lib/ca-certificates/ca-bundle.pem (linked to
>>> /etc/ssl/ca-bundle.pem) installed by the ca-certificates package.
>>>
>>> The OE ca-certificates package does not install such a bundle.
>>> So may be that's the real problem.
> 
>> A look into the ca-certificates recipe shows this:
> 
>>   pkg_postinst_${PN} () {
>>           /usr/sbin/update-ca-certificates
>>   }
> 
>> I think this is to late if another package needs the result from
>> this call in its configuration stage.
> 
> That's missing a $D check as well, since in that form it can't be run
> off-line

Can please someone else have a look at this? I'm not familiar with this stuff (and BTW i'm not the maintainer of epiphany and/or gnome).

Steffen

Re: testing branch 2010-10-08

[Cliff Brake]

On Sat, Oct 9, 2010 at 7:30 PM, Graham Gower <graham.gower@gmail.com> wrote:
> On 9 October 2010 17:40, Frans Meulenbroeks <fransmeulenbroeks@gmail.com> wrote:
>> Thanks for doing this Cliff.
>> I've got my 6 builds already scheduled on our autobuilder.I will
>> report back on the wiki.
>>
>> One small remark: timewise, I would prefer creation of the branch of
>> your friday morning. For us europeans the weekend is already started.
>> Doing it a little bit earlier allows people like Steffen and me to
>> kick it off before we go home.
>
> How about Thursday, that way us UTC+9:30ians (+1hour daylight
> savings!) can kick it off before our weekend :)
>
>> Alternately it might help if we can define a time before which the
>> branch is made. That would allow a cron jub or so to kick things off.
>
> I like this idea too.

I'll plan on merging testing-next at end of business day Thursday
(EST), 9:00PM GMT.  That way people can start their builds any time on
Fri and we still have most of Thurs to report issues, etc.

Thanks,
Cliff

-- 
=================
http://bec-systems.com

Re: testing branch 2010-10-08

[Frans Meulenbroeks]

2010/10/12 Cliff Brake <cliff.brake@gmail.com>:
> On Sat, Oct 9, 2010 at 7:30 PM, Graham Gower <graham.gower@gmail.com> wrote:
>> On 9 October 2010 17:40, Frans Meulenbroeks <fransmeulenbroeks@gmail.com> wrote:
>>> Thanks for doing this Cliff.
>>> I've got my 6 builds already scheduled on our autobuilder.I will
>>> report back on the wiki.
>>>
>>> One small remark: timewise, I would prefer creation of the branch of
>>> your friday morning. For us europeans the weekend is already started.
>>> Doing it a little bit earlier allows people like Steffen and me to
>>> kick it off before we go home.
>>
>> How about Thursday, that way us UTC+9:30ians (+1hour daylight
>> savings!) can kick it off before our weekend :)
>>
>>> Alternately it might help if we can define a time before which the
>>> branch is made. That would allow a cron jub or so to kick things off.
>>
>> I like this idea too.
>
> I'll plan on merging testing-next at end of business day Thursday
> (EST), 9:00PM GMT.  That way people can start their builds any time on
> Fri and we still have most of Thurs to report issues, etc.
>

Cool, Thanks!

Re: testing branch 2010-10-08

[Petr Štetiar]

Cliff Brake <cliff.brake@gmail.com> [2010-10-08 17:35:54]:

> master branch has been merged to testing-next and is ready for clean builds.

Hi,

I've build the testing-next 2010-10-08 for ts72xx machine, distro
angstrom-2010.x on Ubuntu 10.04.1 32-bit. I've updated the testing table on
the wiki. Is it enough or am I missing some additional step? Thanks.

-- ynezz

Re: testing branch 2010-10-08

[Khem Raj]

On Wed, Oct 13, 2010 at 9:36 AM, Petr Štetiar <ynezz@true.cz> wrote:
> Cliff Brake <cliff.brake@gmail.com> [2010-10-08 17:35:54]:
>
>> master branch has been merged to testing-next and is ready for clean builds.
>
> Hi,
>
> I've build the testing-next 2010-10-08 for ts72xx machine, distro
> angstrom-2010.x on Ubuntu 10.04.1 32-bit. I've updated the testing table on
> the wiki. Is it enough or am I missing some additional step? Thanks.

thank you. I think if you update the table and all its fields its good enough.

>
> -- ynezz
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>