[RFC] New stable branch (and maybe a new process?)

[RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

I want to gather some feedback what people are feeling about a new stable
branch. It feels like the right time to do it soon. We have so many big changes
in that are not likely to get into stable/2009 like:

- New staging
- BBCLASSEXTEND
- Checksums in recipes
- Faster BitBake
- Enhanced java support with openjdk/icedtea
- And many I forgot...

With the latest fixes for the file:// problem the build feels stable at last for
me. I have not been involved with the latest stable branch so this is really
only a RFC from my side. Please let me know if you see big issues for such a
move.

Open questions:
---------------
- Do we feel comfortable with doing a new branch now?
- What version of bitbake we would like to have for this?
- What boards/images/distros will be supported in stable/2010?

I my be biased here regarding my work for BugLabs, but I feel me diverged that
much from stable/2009 that it gets problematic to get changes in and syncing
them with .dev. As well the overall build experience on .dev is pretty good the
last weeks. The file:// problems aside.

Another idea we just discussed on IRC was to have a model somehow related to
what debian is using. An ongoing flow of patches from .dev to a testing branch.
Perhaps automated with buildbot to move into testing when some certain builds
have passed with this commit. And also an _ongoing_ stream well tested patches
from testing into stable which eventually makes a time-based release every X
months form this ond move on.

That all needs to be filled with more details, but it could be a start. Let me
know what you think.

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Frans Meulenbroeks]

2010/7/9 Stefan Schmidt <stefan@datenfreihafen.org>:
> Hello.
>
> I want to gather some feedback what people are feeling about a new stable
> branch. It feels like the right time to do it soon. We have so many big changes
> in that are not likely to get into stable/2009 like:
>
> - New staging
> - BBCLASSEXTEND
> - Checksums in recipes
> - Faster BitBake
> - Enhanced java support with openjdk/icedtea
> - And many I forgot...
>
> With the latest fixes for the file:// problem the build feels stable at last for
> me. I have not been involved with the latest stable branch so this is really
> only a RFC from my side. Please let me know if you see big issues for such a
> move.
>
> Open questions:
> ---------------
> - Do we feel comfortable with doing a new branch now?

I'd say we should wailt until bitbake 1.10 is officially out (I assume
from your previous "Faster bitbake" that you want to move to 1.10)

> - What version of bitbake we would like to have for this?

I'd say 1.10

> - What boards/images/distros will be supported in stable/2010?
>
> I my be biased here regarding my work for BugLabs, but I feel me diverged that
> much from stable/2009 that it gets problematic to get changes in and syncing
> them with .dev. As well the overall build experience on .dev is pretty good the
> last weeks. The file:// problems aside.
>
> Another idea we just discussed on IRC was to have a model somehow related to
> what debian is using. An ongoing flow of patches from .dev to a testing branch.
> Perhaps automated with buildbot to move into testing when some certain builds
> have passed with this commit. And also an _ongoing_ stream well tested patches
> from testing into stable which eventually makes a time-based release every X
> months form this ond move on.
>
> That all needs to be filled with more details, but it could be a start. Let me
> know what you think.
>
> regards
> Stefan Schmidt
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Sat, 2010-07-10 at 08:51, Frans Meulenbroeks wrote:
> 2010/7/9 Stefan Schmidt <stefan@datenfreihafen.org>:
> >
> > Open questions:
> > ---------------
> > - Do we feel comfortable with doing a new branch now?
> 
> I'd say we should wailt until bitbake 1.10 is officially out (I assume
> from your previous "Faster bitbake" that you want to move to 1.10)

Agreed. The only problem we had recently was about file:// handling. Everything
else seems to be pretty stable on the 1.10 branch. Could one of the bitbake devs
comment on the question if there is a 1.10 release planned?

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Fri, 2010-07-09 at 22:40, Stefan Schmidt wrote:
> 
> That all needs to be filled with more details, but it could be a start. Let me
> know what you think.

I have to say I expected more replies. Some blames here and there and maybe a
discussion. To hot these days? :)

More serious, I think it would be a good time now to discuss this. What
experience did the stable maintainer made with stable/2009 so far? Is it worth
doing it again? Maybe change the process?

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Fri, 2010-07-09 at 22:40, Stefan Schmidt wrote:
> 
> I want to gather some feedback what people are feeling about a new stable
> branch. It feels like the right time to do it soon. We have so many big changes
> in that are not likely to get into stable/2009 like:

From the silence on this topic I gather that the interest to create a new branch
or change the process is quite low.

Lets see how painful it will be to get all our overlay changes into stable/2009
then. Some review/acks on the outstanding patches would be a good start here ;)

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Koen Kooi]

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

On 20-07-10 09:01, Stefan Schmidt wrote:
> Hello.
> 
> On Fri, 2010-07-09 at 22:40, Stefan Schmidt wrote:
>>
>> I want to gather some feedback what people are feeling about a new stable
>> branch. It feels like the right time to do it soon. We have so many big changes
>> in that are not likely to get into stable/2009 like:
> 
> From the silence on this topic I gather that the interest to create a new branch
> or change the process is quite low.

I have an interest in a new stable branch, but no interest in changing
the process.
The biggest problem is the .dev is halfway into big changes, e.g.
halfway into new-style staging, halfway into packaged-staging rework,
halfway into nativesdk, etc.

regards,

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

iD8DBQFMRVnWMkyGM64RGpERAoerAJ9u8eZiJ2tYl/fn6md8Iy++pq/hMgCfZDYe
dkRT0fmRS9WMdNRqf2GBNcY=
=Zcrp
-----END PGP SIGNATURE-----

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Tue, 2010-07-20 at 10:09, Koen Kooi wrote:
> On 20-07-10 09:01, Stefan Schmidt wrote:
> > 
> > From the silence on this topic I gather that the interest to create a new branch
> > or change the process is quite low.
> 
> I have an interest in a new stable branch, but no interest in changing
> the process.

Fair enough. Lets handle the issues seperately for now.

> The biggest problem is the .dev is halfway into big changes, e.g.
> halfway into new-style staging,

Agreed. While I think we are more then halfway through with this. Recipes are
left but most of the active maintained are already done. NO numbers but my
feeling here.

> halfway into packaged-staging rework, halfway into nativesdk, etc.

These both are ongoing for quite some time already. I have to admit that I lost
track where they stand and what is still outstanding.

So your suggestion is to wait until all three are done?

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Chris Simmonds]

On 20/07/10 08:01, Stefan Schmidt wrote:
> Hello.
>
> On Fri, 2010-07-09 at 22:40, Stefan Schmidt wrote:
>>
>> I want to gather some feedback what people are feeling about a new stable
>> branch. It feels like the right time to do it soon. We have so many big changes
>> in that are not likely to get into stable/2009 like:
>
>> From the silence on this topic I gather that the interest to create a new branch
> or change the process is quite low.
>

I would like to vote for a 2010 stable branch. There is a lot of good 
stuff in the dev branch that I would like to use in live projects.

Bye for now,
Chris.

Re: [RFC] New stable branch (and maybe a new process?)

[Robert P. J. Day]

On Tue, 20 Jul 2010, Chris Simmonds wrote:

> On 20/07/10 08:01, Stefan Schmidt wrote:
> > Hello.
> >
> > On Fri, 2010-07-09 at 22:40, Stefan Schmidt wrote:
> > >
> > > I want to gather some feedback what people are feeling about a new stable
> > > branch. It feels like the right time to do it soon. We have so many big
> > > changes
> > > in that are not likely to get into stable/2009 like:
> >
> > > From the silence on this topic I gather that the interest to create a new
> > > branch
> > or change the process is quite low.
> >
>
> I would like to vote for a 2010 stable branch. There is a lot of
> good stuff in the dev branch that I would like to use in live
> projects.

  i second that, with the proviso that the primary focus for that
new stable branch be on resolving any unbuildable packages for major
distros.  as i've whined about before, there always seems to be a
package or two from the dev branch that doesn't build on my standard
64-bit ubuntu system (last time i checked, it was abiword).  i just
did a git pull on the dev branch and am checking again to see what
fails.

  not to belabour the point, but if a new branch is created and tagged
as "stable," it should build for anyone using any of the common
distros.

rday

-- 

========================================================================
Robert P. J. Day                               Waterloo, Ontario, CANADA

        Top-notch, inexpensive online Linux/OSS/kernel courses
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: [RFC] New stable branch (and maybe a new process?)

[Jaap de Jong]

stable/2010?
Yes, please!!

Bye,
Jaap

On 07/20/2010 10:43 AM, Chris Simmonds wrote:
> On 20/07/10 08:01, Stefan Schmidt wrote:
>> Hello.
>>
>> On Fri, 2010-07-09 at 22:40, Stefan Schmidt wrote:
>>>
>>> I want to gather some feedback what people are feeling about a new 
>>> stable
>>> branch. It feels like the right time to do it soon. We have so many 
>>> big changes
>>> in that are not likely to get into stable/2009 like:
>>
>>> From the silence on this topic I gather that the interest to create 
>>> a new branch
>> or change the process is quite low.
>>
>
> I would like to vote for a 2010 stable branch. There is a lot of good 
> stuff in the dev branch that I would like to use in live projects.
>
> Bye for now,
> Chris.
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: [RFC] New stable branch (and maybe a new process?)

[Christophe Aeschlimann]

Hi,

On 20.07.2010 12:12, Jaap de Jong wrote:
> stable/2010?
> Yes, please!!

I second that :)

>
> Bye,
> Jaap
>
> On 07/20/2010 10:43 AM, Chris Simmonds wrote:
>> On 20/07/10 08:01, Stefan Schmidt wrote:
>>> Hello.
>>>
>>> On Fri, 2010-07-09 at 22:40, Stefan Schmidt wrote:
>>>>
>>>> I want to gather some feedback what people are feeling about a new
>>>> stable
>>>> branch. It feels like the right time to do it soon. We have so many
>>>> big changes
>>>> in that are not likely to get into stable/2009 like:
>>>
>>>> From the silence on this topic I gather that the interest to create
>>>> a new branch
>>> or change the process is quite low.
>>>
>>
>> I would like to vote for a 2010 stable branch. There is a lot of good
>> stuff in the dev branch that I would like to use in live projects.
>>
>> Bye for now,
>> Chris.
>>
>> _______________________________________________
>> 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
>


-- 
Christophe Aeschlimann

Embedded Software Engineer

Advanced Communications Networks S.A.

Rue du Puits-Godet 8a
2000 Neuchâtel, Switzerland

Tél. +41 32 724 74 31

c.aeschlimann@acn-group.ch

Re: [RFC] New stable branch (and maybe a new process?)

[Detlef Vollmann]

On 07/20/10 10:40, Stefan Schmidt wrote:
> On Tue, 2010-07-20 at 10:09, Koen Kooi wrote:
>> The biggest problem is the .dev is halfway into big changes, e.g.
>> halfway into new-style staging,
> 
> Agreed. While I think we are more then halfway through with this. Recipes are
> left but most of the active maintained are already done. NO numbers but my
> feeling here.
> 
>> halfway into packaged-staging rework, halfway into nativesdk, etc.
> 
> These both are ongoing for quite some time already. I have to admit that I lost
> track where they stand and what is still outstanding.
> 
> So your suggestion is to wait until all three are done?
That would be my suggestion.
What "done" exactly means is a separate issue.

   Detlef

Re: [RFC] New stable branch (and maybe a new process?)

[Philip Balister]

On 07/20/2010 04:40 AM, Stefan Schmidt wrote:
> Hello.
>
> On Tue, 2010-07-20 at 10:09, Koen Kooi wrote:
>> On 20-07-10 09:01, Stefan Schmidt wrote:
>>>
>>>  From the silence on this topic I gather that the interest to create a new branch
>>> or change the process is quite low.
>>
>> I have an interest in a new stable branch, but no interest in changing
>> the process.
>
> Fair enough. Lets handle the issues seperately for now.
>
>> The biggest problem is the .dev is halfway into big changes, e.g.
>> halfway into new-style staging,
>
> Agreed. While I think we are more then halfway through with this. Recipes are
> left but most of the active maintained are already done. NO numbers but my
> feeling here.
>
>> halfway into packaged-staging rework, halfway into nativesdk, etc.
>
> These both are ongoing for quite some time already. I have to admit that I lost
> track where they stand and what is still outstanding.
>
> So your suggestion is to wait until all three are done?

We should identify the work in progress in dev and decide which ones we 
should finish before creating a new staging. There is no sense making a 
new stable branch and then creating a major diversion from dev that will 
make maintaining both in parallel for as long as possible difficult.

Philip

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Tue, 2010-07-20 at 09:10, Philip Balister wrote:
> On 07/20/2010 04:40 AM, Stefan Schmidt wrote:
> >On Tue, 2010-07-20 at 10:09, Koen Kooi wrote:
> >
> >>The biggest problem is the .dev is halfway into big changes, e.g.
> >>halfway into new-style staging,
> >
> >Agreed. While I think we are more then halfway through with this. Recipes are
> >left but most of the active maintained are already done. NO numbers but my
> >feeling here.
> >
> >>halfway into packaged-staging rework, halfway into nativesdk, etc.
> >
> >These both are ongoing for quite some time already. I have to admit that I lost
> >track where they stand and what is still outstanding.
> >
> >So your suggestion is to wait until all three are done?
> 
> We should identify the work in progress in dev and decide which ones
> we should finish before creating a new staging. There is no sense
> making a new stable branch and then creating a major diversion from
> dev that will make maintaining both in parallel for as long as
> possible difficult.

Does my mail read as if I wanted to rush this in? If yes, that was not my
intend.

I'm interested in new stable branch. I had a moment where I hoped it would be
ready in time for our upcoming BUG2.0 release, but this moment is over already.
Neverless I want to see it happened and it should not be rushed in.

The question was completely honest. Is the idea to wait until all three are
finished (for every recipe in tree)?

For the new staging I could imagine that we could get close. Of course minus all
the recipes in tree nobody maintains. A problem we always have and which does
not go away in a stable snapshot.

For packaged staging it may come true as well, but the nativesdk part I see
mostly arguing back and forth. So if the people listed as stable maintainers
come to the conclusion that we should wait for all three that is fine with me.
Maybe a roadmap what you guys want to have ready before branching of would help
others then.

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Koen Kooi]

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

On 20-07-10 10:40, Stefan Schmidt wrote:
> Hello.
> 
> On Tue, 2010-07-20 at 10:09, Koen Kooi wrote:
>> On 20-07-10 09:01, Stefan Schmidt wrote:
>>>
>>> From the silence on this topic I gather that the interest to create a new branch
>>> or change the process is quite low.
>>
>> I have an interest in a new stable branch, but no interest in changing
>> the process.
> 
> Fair enough. Lets handle the issues seperately for now.
> 
>> The biggest problem is the .dev is halfway into big changes, e.g.
>> halfway into new-style staging,
> 
> Agreed. While I think we are more then halfway through with this. Recipes are
> left but most of the active maintained are already done. NO numbers but my
> feeling here.
> 
>> halfway into packaged-staging rework, halfway into nativesdk, etc.
> 
> These both are ongoing for quite some time already. I have to admit that I lost
> track where they stand and what is still outstanding.
> 
> So your suggestion is to wait until all three are done?

'Wait' is maybe the wrong word. I would try to do a call to arms to get
more people interested into getting this done. I had hoped that the TSC
would be more vocal about this, but we must do with the cards dealt.

And of course the number of people screaming for a stable branch is an
order of magnitude larger than the number of people that end up actually
supporting it.

regards,

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

iD8DBQFMRdUmMkyGM64RGpERAmlcAJ0TbmzdYV6Z8p+FdtCP1jwXWOdIowCfbEJ9
8Zyd7nsYgEZZfj2daMq+EdI=
=gEzx
-----END PGP SIGNATURE-----

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
> On 20-07-10 10:40, Stefan Schmidt wrote:
> > 
> > So your suggestion is to wait until all three are done?
> 
> 'Wait' is maybe the wrong word. I would try to do a call to arms to get
> more people interested into getting this done.

Wasn't there a list produced with all recipes that used legacy staging? I swear
I have seen something like this but I can't find it in my recent builds.

I will have an eye on the legacy staging things and will read into packaged
staging. Nativesdk I would leave for others.

> And of course the number of people screaming for a stable branch is an
> order of magnitude larger than the number of people that end up actually
> supporting it.

Sure, that is no surprise. :)

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Koen Kooi]

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

On 20-07-10 19:34, Stefan Schmidt wrote:
> Hello.
> 
> On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
>> On 20-07-10 10:40, Stefan Schmidt wrote:
>>>
>>> So your suggestion is to wait until all three are done?
>>
>> 'Wait' is maybe the wrong word. I would try to do a call to arms to get
>> more people interested into getting this done.
> 
> Wasn't there a list produced with all recipes that used legacy staging? I swear
> I have seen something like this but I can't find it in my recent builds.

In your TMPDIR:

koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq | wc -l
53

It includes beasts like perl and python.

regards,

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

iD8DBQFMRe4CMkyGM64RGpERAv89AJ9jHvcsxzDIQAYNIHbJR1IWV3PlDACeNYn2
qVizRtaDD5OWWKuSSRb5RcY=
=Q4uM
-----END PGP SIGNATURE-----

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Tue, 2010-07-20 at 20:42, Koen Kooi wrote:
> 
> On 20-07-10 19:34, Stefan Schmidt wrote:
> > 
> > Wasn't there a list produced with all recipes that used legacy staging? I swear
> > I have seen something like this but I can't find it in my recent builds.
> 
> In your TMPDIR:
> 
> koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq | wc -l
> 53

Ha, there it was. :)

> It includes beasts like perl and python.

Hmm, that would be tricky. But there are some more easier ones that could get
nucked before.

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Esben Haabendal]

On Tue, Jul 20, 2010 at 7:34 PM, Stefan Schmidt
<stefan@datenfreihafen.org> wrote:
> On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
>> And of course the number of people screaming for a stable branch is an
>> order of magnitude larger than the number of people that end up actually
>> supporting it.
>
> Sure, that is no surprise. :)

It might be surprising to some, but some of those not actively
supporting the "stable" branch might choose not to do so because the
chosen stable process does not fit their needs.
Most likely, all people contributing to stable is mostly happy with
how it is handled.

/Esben

Re: [RFC] New stable branch (and maybe a new process?)

[Tom Rini]

Koen Kooi wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On 20-07-10 19:34, Stefan Schmidt wrote:
>> Hello.
>>
>> On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
>>> On 20-07-10 10:40, Stefan Schmidt wrote:
>>>> So your suggestion is to wait until all three are done?
>>> 'Wait' is maybe the wrong word. I would try to do a call to arms to get
>>> more people interested into getting this done.
>> Wasn't there a list produced with all recipes that used legacy staging? I swear
>> I have seen something like this but I can't find it in my recent builds.
> 
> In your TMPDIR:
> 
> koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq | wc -l
> 53
> 
> It includes beasts like perl and python.

perl yes, python no.  Overall point, yes.

-- 
Tom Rini
Mentor Graphics Corporation

Re: [RFC] New stable branch (and maybe a new process?)

[Stefan Schmidt]

Hello.

On Tue, 2010-07-20 at 21:07, Esben Haabendal wrote:
> On Tue, Jul 20, 2010 at 7:34 PM, Stefan Schmidt
> <stefan@datenfreihafen.org> wrote:
> > On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
> >> And of course the number of people screaming for a stable branch is an
> >> order of magnitude larger than the number of people that end up actually
> >> supporting it.
> >
> > Sure, that is no surprise. :)
> 
> It might be surprising to some, but some of those not actively
> supporting the "stable" branch might choose not to do so because the
> chosen stable process does not fit their needs.
> Most likely, all people contributing to stable is mostly happy with
> how it is handled.

People who add a "me, too" for a new stable branch seem to think that they fit
their needs.

What does not fit your needs with the current stable handling?

I would like to have a more continous flow of changes from dev over some testing
branch to stable. As indicated in my initial mail. That would people working on
this idea and keeping it alive though. Given that no one seemed to be interested
but man seem to be interested to get a new stable/2010 branch the same way we
had for 2009 it seems the most practical way to get one at all, no?

Other models will only work if enough support from the active contributors is
available. Without this even the best idea might die.

regards
Stefan Schmidt

Re: [RFC] New stable branch (and maybe a new process?)

[Koen Kooi]

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

On 20-07-10 21:16, Tom Rini wrote:
> Koen Kooi wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On 20-07-10 19:34, Stefan Schmidt wrote:
>>> Hello.
>>>
>>> On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
>>>> On 20-07-10 10:40, Stefan Schmidt wrote:
>>>>> So your suggestion is to wait until all three are done?
>>>> 'Wait' is maybe the wrong word. I would try to do a call to arms to get
>>>> more people interested into getting this done.
>>> Wasn't there a list produced with all recipes that used legacy
>>> staging? I swear
>>> I have seen something like this but I can't find it in my recent builds.
>>
>> In your TMPDIR:
>>
>> koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq |
>> wc -l
>> 53
>>
>> It includes beasts like perl and python.
> 
> perl yes, python no.  Overall point, yes.
> 

Mine lists:

/OE/org.openembedded.dev/recipes/perl/perl_5.8.8.bb
/OE/org.openembedded.dev/recipes/python/python_2.6.4.bb

And that's from a rebuild from scratch last week.

regards,

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

iD8DBQFMRg3hMkyGM64RGpERAu2OAKCsIiUFeNEVUZkqBKLP9TKO9yW3eACfQ2rx
ZTxsG0zc2k5hnjkwwDVwiAo=
=Yjad
-----END PGP SIGNATURE-----

Re: [RFC] New stable branch (and maybe a new process?)

[Frans Meulenbroeks]

2010/7/20 Koen Kooi <k.kooi@student.utwente.nl>

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 20-07-10 19:34, Stefan Schmidt wrote:
> > Hello.
> >
> > On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
> >> On 20-07-10 10:40, Stefan Schmidt wrote:
> >>>
> >>> So your suggestion is to wait until all three are done?
> >>
> >> 'Wait' is maybe the wrong word. I would try to do a call to arms to get
> >> more people interested into getting this done.
> >
> > Wasn't there a list produced with all recipes that used legacy staging? I
> swear
> > I have seen something like this but I can't find it in my recent builds.
>
> In your TMPDIR:
>
> koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq | wc
> -l
> 53
>

If I understand things correctly these are only the recipes that you have
actually build that have legacy-staging (my current file lists only 1:
libtool).
However there are way more recipes around that use legacy staging.
Not fully sure if this is a right measure but there are 1140 recipes in my
tree that have a do_stage word in it. 21 of these are things like
do_stage_append, the other 1119 are having do_stage as a word.

Grep output from my tree:

frans@linux-suse:~/oe/openembedded/recipes> grep -l do_stage */*.bb | wc -l
1140
frans@linux-suse:~/oe/openembedded/recipes> grep -l -w do_stage */*.bb | wc
-l
1119

I suspect all of these need fixing (or removal).

Taking alsa/alsa-lib as an example (as this one was near the top of the list
if you do a grep.
There are 9 versions in the tree:
alsa-lib_1.0.11.bb
alsa-lib_1.0.13.bb
alsa-lib_1.0.14.bb
alsa-lib_1.0.15.bb
alsa-lib_1.0.17.bb
alsa-lib_1.0.18.bb
alsa-lib_1.0.19.bb
alsa-lib_1.0.20.bb
alsa-lib_1.0.23.bb
The last two do not have a do_stage. The first 7 do.
If we want to abandon legacy staging the first 7 should be fixed too (which
implies reworking the recpie and checking if it builds and installs
properly).
Alternate of course could be to remove those recipes (which could be done
minus .15 which is pinned:
conf/distro/include/preferred-om-2008-versions.inc:PREFERRED_VERSION_alsa-lib
?= "1.0.15"

As I wrote this is just an example. There seem to be 1119 files to go

Frans

PS: and apologies for digressing the discussion.


> It includes beasts like perl and python.
>
> regards,
>
> KOen
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.5 (Darwin)
>
> iD8DBQFMRe4CMkyGM64RGpERAv89AJ9jHvcsxzDIQAYNIHbJR1IWV3PlDACeNYn2
> qVizRtaDD5OWWKuSSRb5RcY=
> =Q4uM
> -----END PGP SIGNATURE-----
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [RFC] New stable branch (and maybe a new process?)

[Richard Purdie]

On Tue, 2010-07-20 at 21:04 +0200, Stefan Schmidt wrote:
> Hello.
> 
> On Tue, 2010-07-20 at 20:42, Koen Kooi wrote:
> > 
> > On 20-07-10 19:34, Stefan Schmidt wrote:
> > > 
> > > Wasn't there a list produced with all recipes that used legacy staging? I swear
> > > I have seen something like this but I can't find it in my recent builds.
> > 
> > In your TMPDIR:
> > 
> > koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq | wc -l
> > 53
> 
> Ha, there it was. :)
> 
> > It includes beasts like perl and python.
> 
> Hmm, that would be tricky. But there are some more easier ones that could get
> nucked before.

Poky has removed legacy staging for perl and python so it is possible
and there is an example out there...

Cheers,

Richard

Re: [RFC] New stable branch (and maybe a new process?)

[Tom Rini]

Koen Kooi wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On 20-07-10 21:16, Tom Rini wrote:
>> Koen Kooi wrote:
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>>
>>> On 20-07-10 19:34, Stefan Schmidt wrote:
>>>> Hello.
>>>>
>>>> On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
>>>>> On 20-07-10 10:40, Stefan Schmidt wrote:
>>>>>> So your suggestion is to wait until all three are done?
>>>>> 'Wait' is maybe the wrong word. I would try to do a call to arms to get
>>>>> more people interested into getting this done.
>>>> Wasn't there a list produced with all recipes that used legacy
>>>> staging? I swear
>>>> I have seen something like this but I can't find it in my recent builds.
>>> In your TMPDIR:
>>>
>>> koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq |
>>> wc -l
>>> 53
>>>
>>> It includes beasts like perl and python.
>> perl yes, python no.  Overall point, yes.
>>
> 
> Mine lists:
> 
> /OE/org.openembedded.dev/recipes/perl/perl_5.8.8.bb
> /OE/org.openembedded.dev/recipes/python/python_2.6.4.bb
> 
> And that's from a rebuild from scratch last week.

Going off list... 
http://gitweb.openembedded.net/?p=openembedded.git&a=commit;h=a753e680011a9843e0cd0ee928fd508667445d00 
and I just did a -c clean and rebuild.

-- 
Tom Rini
Mentor Graphics Corporation

Re: [RFC] New stable branch (and maybe a new process?)

[Tom Rini]

Tom Rini wrote:
> Koen Kooi wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On 20-07-10 21:16, Tom Rini wrote:
>>> Koen Kooi wrote:
>>>> -----BEGIN PGP SIGNED MESSAGE-----
>>>> Hash: SHA1
>>>>
>>>> On 20-07-10 19:34, Stefan Schmidt wrote:
>>>>> Hello.
>>>>>
>>>>> On Tue, 2010-07-20 at 18:56, Koen Kooi wrote:
>>>>>> On 20-07-10 10:40, Stefan Schmidt wrote:
>>>>>>> So your suggestion is to wait until all three are done?
>>>>>> 'Wait' is maybe the wrong word. I would try to do a call to arms 
>>>>>> to get
>>>>>> more people interested into getting this done.
>>>>> Wasn't there a list produced with all recipes that used legacy
>>>>> staging? I swear
>>>>> I have seen something like this but I can't find it in my recent 
>>>>> builds.
>>>> In your TMPDIR:
>>>>
>>>> koen@dominion:/OE/angstrom-dev$ cat legacy-staging.log | sort | uniq |
>>>> wc -l
>>>> 53
>>>>
>>>> It includes beasts like perl and python.
>>> perl yes, python no.  Overall point, yes.
>>>
>>
>> Mine lists:
>>
>> /OE/org.openembedded.dev/recipes/perl/perl_5.8.8.bb
>> /OE/org.openembedded.dev/recipes/python/python_2.6.4.bb
>>
>> And that's from a rebuild from scratch last week.
> 
> Going off list... 
> http://gitweb.openembedded.net/?p=openembedded.git&a=commit;h=a753e680011a9843e0cd0ee928fd508667445d00 
> and I just did a -c clean and rebuild.

Welp, I fail today.  Sorry all.

-- 
Tom Rini
Mentor Graphics Corporation

Re: [RFC] New stable branch (and maybe a new process?)

[Esben Haabendal]

On Tue, Jul 20, 2010 at 9:22 PM, Stefan Schmidt
<stefan@datenfreihafen.org> wrote:

> What does not fit your needs with the current stable handling?

It started out with a rather continous flow of changed from dev,
causing it to be almost as unstable as dev. Now stable/2009 looks
mostly dead.

What I need is a really _STABLE_ branch, but still maintained, on
which it is possible to base products on for customers which do not
want to hear about problems with the chosen upstream build system.
I realize that I am rather alone with my wishes for stable, so I just
rather silently express my wishes here for the record.

> I would like to have a more continous flow of changes from dev over some testing
> branch to stable. As indicated in my initial mail. That would people working on
> this idea and keeping it alive though. Given that no one seemed to be interested
> but man seem to be interested to get a new stable/2010 branch the same way we
> had for 2009 it seems the most practical way to get one at all, no?
>
> Other models will only work if enough support from the active contributors is
> available. Without this even the best idea might die.

Exactly, and this is why I don't really want to push my idea on how a
stable branch should work, as I realize I would likely end up being
rather alone maintaining it, which is how it is today with my customer
OE projects anyways.

/Esben
-- 
Esben Haabendal, Senior Software Consultant
DoréDevelopment ApS, Ved Stranden 1, 9560 Hadsund, DK-Denmark
Phone: +45 51 92 53 93, E-mail: eha@doredevelopment.dk
WWW: http://www.doredevelopment.dk

Re: [RFC] New stable branch (and maybe a new process?)

[Frans Meulenbroeks]

2010/7/21 Esben Haabendal <esbenhaabendal@gmail.com>

> On Tue, Jul 20, 2010 at 9:22 PM, Stefan Schmidt
> <stefan@datenfreihafen.org> wrote:
>
> > What does not fit your needs with the current stable handling?
>
> It started out with a rather continous flow of changed from dev,
> causing it to be almost as unstable as dev. Now stable/2009 looks
> mostly dead.
>
> What I need is a really _STABLE_ branch, but still maintained, on
> which it is possible to base products on for customers which do not
> want to hear about problems with the chosen upstream build system.
> I realize that I am rather alone with my wishes for stable, so I just
> rather silently express my wishes here for the record.
>

You can create your own. That'll give you all the control.
Alternately you just could decide to freeze whatever version you are using
and only cherry pick some bugfixes.
If you are doing end user products it may be a good plan to do a freeze at
some point anyway, otherwise you're shooting at a moving target.

Frans (who understands your desire)

>
> > I would like to have a more continous flow of changes from dev over some
> testing
> > branch to stable. As indicated in my initial mail. That would people
> working on
> > this idea and keeping it alive though. Given that no one seemed to be
> interested
> > but man seem to be interested to get a new stable/2010 branch the same
> way we
> > had for 2009 it seems the most practical way to get one at all, no?
> >
> > Other models will only work if enough support from the active
> contributors is
> > available. Without this even the best idea might die.
>
> Exactly, and this is why I don't really want to push my idea on how a
> stable branch should work, as I realize I would likely end up being
> rather alone maintaining it, which is how it is today with my customer
> OE projects anyways.
>
> /Esben
> --
> Esben Haabendal, Senior Software Consultant
> DoréDevelopment ApS, Ved Stranden 1, 9560 Hadsund, DK-Denmark
> Phone: +45 51 92 53 93, E-mail: eha@doredevelopment.dk
> WWW: http://www.doredevelopment.dk
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [RFC] New stable branch (and maybe a new process?)

[Khem Raj]

On Wed, Jul 21, 2010 at 12:13 AM, Frans Meulenbroeks
<fransmeulenbroeks@gmail.com> wrote:
> 2010/7/21 Esben Haabendal <esbenhaabendal@gmail.com>
>
>> On Tue, Jul 20, 2010 at 9:22 PM, Stefan Schmidt
>> <stefan@datenfreihafen.org> wrote:
>>
>> > What does not fit your needs with the current stable handling?
>>
>> It started out with a rather continous flow of changed from dev,
>> causing it to be almost as unstable as dev. Now stable/2009 looks
>> mostly dead.
>>

I think  a time based release will be really nice for OE for improving its
consumption in production deployment, I am willing to help in this
How often we want to release remains to be decided.
We don't have to do huge backports and wait for a lot of features
to go in for cutting out a stable release.

-Khem