Re: [oe-commits] Khem Raj : binutils-cross: Sync with oe-core

Re: [oe-commits] Khem Raj : binutils-cross: Sync with oe-core

[Andrea Adami]

On Sat, Oct 15, 2011 at 8:51 PM, <git@git.openembedded.org> wrote:

> Module: meta-openembedded.git
> Branch: master
> Commit: 92b02b7209e426a70cc5626e7bdbc82052e01ac5
> URL:
> http://git.openembedded.org/?p=meta-openembedded.git&a=commit;h=92b02b7209e426a70cc5626e7bdbc82052e01ac5
>
> Author: Khem Raj <raj.khem@gmail.com>
> Date:   Sun Oct 16 01:28:32 2011 +0000
>
> binutils-cross: Sync with oe-core
>
> Now we have own copy of binutils-cross in meta-oe
>

I can't unfortunately find the discussion which supposedly provided a valid
reason for binutils-2.20 in meta-oe while binutils-2.21 is in oe-core.
I'm against that kind of policy for meta-oe, particularly when it's about
toolchain.

we need to sync with oe-core periodically
>
That's just one of the bad effects :/


>
> Signed-off-by: Khem Raj <raj.khem@gmail.com>
> Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
>
>
my 2c

Andrea

Re: [oe-commits] Khem Raj : binutils-cross: Sync with oe-core

[Khem Raj]

On Sun, Oct 16, 2011 at 2:37 PM, Andrea Adami <andrea.adami@gmail.com> wrote:
> On Sat, Oct 15, 2011 at 8:51 PM, <git@git.openembedded.org> wrote:
>
>> Module: meta-openembedded.git
>> Branch: master
>> Commit: 92b02b7209e426a70cc5626e7bdbc82052e01ac5
>> URL:
>> http://git.openembedded.org/?p=meta-openembedded.git&a=commit;h=92b02b7209e426a70cc5626e7bdbc82052e01ac5
>>
>> Author: Khem Raj <raj.khem@gmail.com>
>> Date:   Sun Oct 16 01:28:32 2011 +0000
>>
>> binutils-cross: Sync with oe-core
>>
>> Now we have own copy of binutils-cross in meta-oe
>>
>
> I can't unfortunately find the discussion which supposedly provided a valid
> reason for binutils-2.20 in meta-oe while binutils-2.21 is in oe-core.
> I'm against that kind of policy for meta-oe, particularly when it's about
> toolchain.

oe-core is deprecating versions that are still used by consumers of
meta-oe (mainly angstrom 2010 release)
since its toolchain and it will also benefit derivatives of angstrom.
Angstrom could also move these elements
into meta-angstrom if we think that its only angstrom specific and no
other consumers of oe-core needs them
layer maintainers make judgement call which they think should benefit
wider community

similar but not for binutils here is a relevant thread

http://lists.linuxtogo.org/pipermail/openembedded-core/2011-October/010700.html

I think angstrom needs older binutils to support older kernels.

>
> we need to sync with oe-core periodically
>>
> That's just one of the bad effects :/
>
>
>>
>> Signed-off-by: Khem Raj <raj.khem@gmail.com>
>> Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
>>
>>
> my 2c
>
> Andrea
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [oe-commits] Khem Raj : binutils-cross: Sync with oe-core

[Frans Meulenbroeks]

2011/10/17 Khem Raj <raj.khem@gmail.com>

> On Sun, Oct 16, 2011 at 2:37 PM, Andrea Adami <andrea.adami@gmail.com>
> wrote:
> > On Sat, Oct 15, 2011 at 8:51 PM, <git@git.openembedded.org> wrote:
> >
> >> Module: meta-openembedded.git
> >> Branch: master
> >> Commit: 92b02b7209e426a70cc5626e7bdbc82052e01ac5
> >> URL:
> >>
> http://git.openembedded.org/?p=meta-openembedded.git&a=commit;h=92b02b7209e426a70cc5626e7bdbc82052e01ac5
> >>
> >> Author: Khem Raj <raj.khem@gmail.com>
> >> Date:   Sun Oct 16 01:28:32 2011 +0000
> >>
> >> binutils-cross: Sync with oe-core
> >>
> >> Now we have own copy of binutils-cross in meta-oe
> >>
> >
> > I can't unfortunately find the discussion which supposedly provided a
> valid
> > reason for binutils-2.20 in meta-oe while binutils-2.21 is in oe-core.
> > I'm against that kind of policy for meta-oe, particularly when it's about
> > toolchain.
>
> oe-core is deprecating versions that are still used by consumers of
> meta-oe (mainly angstrom 2010 release)
> since its toolchain and it will also benefit derivatives of angstrom.
> Angstrom could also move these elements
> into meta-angstrom if we think that its only angstrom specific and no
> other consumers of oe-core needs them
> layer maintainers make judgement call which they think should benefit
> wider community
>
> similar but not for binutils here is a relevant thread
>

I feel if a distro or bsp needs a version of a package that is older than
the oe-core one, it should be stored in the distro or bsp layer.
Or is meta-oe also wanting to keep binutils 18.50? In oe classic this is
used by the nios2 hw (and afaik there is no newer binutils supporting
nios2)? I'd say if some layer needs older versions it is up to them.

What we are lacking here is a policy on purpose/goal of meta-oe and what
goes in (and what not).

Frans

Re: [oe-commits] Khem Raj : binutils-cross: Sync with oe-core

[Khem Raj]

On 10/16/2011 11:35 PM, Frans Meulenbroeks wrote:
> 2011/10/17 Khem Raj<raj.khem@gmail.com>
>
>> On Sun, Oct 16, 2011 at 2:37 PM, Andrea Adami<andrea.adami@gmail.com>
>> wrote:
>>> On Sat, Oct 15, 2011 at 8:51 PM,<git@git.openembedded.org>  wrote:
>>>
>>>> Module: meta-openembedded.git
>>>> Branch: master
>>>> Commit: 92b02b7209e426a70cc5626e7bdbc82052e01ac5
>>>> URL:
>>>>
>> http://git.openembedded.org/?p=meta-openembedded.git&a=commit;h=92b02b7209e426a70cc5626e7bdbc82052e01ac5
>>>>
>>>> Author: Khem Raj<raj.khem@gmail.com>
>>>> Date:   Sun Oct 16 01:28:32 2011 +0000
>>>>
>>>> binutils-cross: Sync with oe-core
>>>>
>>>> Now we have own copy of binutils-cross in meta-oe
>>>>
>>>
>>> I can't unfortunately find the discussion which supposedly provided a
>> valid
>>> reason for binutils-2.20 in meta-oe while binutils-2.21 is in oe-core.
>>> I'm against that kind of policy for meta-oe, particularly when it's about
>>> toolchain.
>>
>> oe-core is deprecating versions that are still used by consumers of
>> meta-oe (mainly angstrom 2010 release)
>> since its toolchain and it will also benefit derivatives of angstrom.
>> Angstrom could also move these elements
>> into meta-angstrom if we think that its only angstrom specific and no
>> other consumers of oe-core needs them
>> layer maintainers make judgement call which they think should benefit
>> wider community
>>
>> similar but not for binutils here is a relevant thread
>>
>
> I feel if a distro or bsp needs a version of a package that is older than
> the oe-core one, it should be stored in the distro or bsp layer.
> Or is meta-oe also wanting to keep binutils 18.50? In oe classic this is
> used by the nios2 hw (and afaik there is no newer binutils supporting
> nios2)? I'd say if some layer needs older versions it is up to them.
>

This is a bit different when a version is retired from oe-core there 
might be more than one bsp using it therefore it would be more 
beneficial to keep it in a common layer

> What we are lacking here is a policy on purpose/goal of meta-oe and what
> goes in (and what not).

This has been discussed quite a bit when we decided to move to layered 
structure. meta-oe infact is an umbrella of layers and each layer can 
have it own development policies.

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

Re: [oe-commits] Khem Raj : binutils-cross: Sync with oe-core

[Frans Meulenbroeks]

2011/10/17 Khem Raj <raj.khem@gmail.com>

> On 10/16/2011 11:35 PM, Frans Meulenbroeks wrote:
>
>>
>>
>> I feel if a distro or bsp needs a version of a package that is older than
>> the oe-core one, it should be stored in the distro or bsp layer.
>> Or is meta-oe also wanting to keep binutils 18.50? In oe classic this is
>> used by the nios2 hw (and afaik there is no newer binutils supporting
>> nios2)? I'd say if some layer needs older versions it is up to them.
>>
>>
> This is a bit different when a version is retired from oe-core there might
> be more than one bsp using it therefore it would be more beneficial to keep
> it in a common layer
>

The risk is that over time this common layer ends up being a repo with
several versions of a recipe, and for some of these we end up having no idea
whether or not they are used.
We've seen in oe classic to what that leads. Here, with layers being
distributed it becomes even more vague what is used and what not.
I'd say if oe core moves forward then a layer that is not ready to move to
that new version should either keep a local copy or pin at a hash of oe-core
before the recipe was removed.
BTW if I recall correctly it was discussed to have an overlap period after
introducing a new version before the old one is deprecated. Not 100% sure if
that was indeed decided that way.

>
>  What we are lacking here is a policy on purpose/goal of meta-oe and what
>> goes in (and what not).
>>
>
> This has been discussed quite a bit when we decided to move to layered
> structure. meta-oe infact is an umbrella of layers and each layer can have
> it own development policies.
>

Guess your are mixing up meta-openembedded (
http://git.openembedded.org/meta-openembedded/tree/) and meta-oe  (
http://git.openembedded.org/meta-openembedded/tree/meta-oe)
(and as far as I know the policys on the common stuff (assuming we see
meta-oe or meta-openembedded as common stuff) is not really documented
afaik).

Btw I feel it does not really help to have some gnome stuff in
meta-oe/recipes-gnome and other in meta-gnome.

Frans