All of lore.kernel.org
 help / color / mirror / Atom feed
* Core image recipes
@ 2011-08-26  9:47 Paul Eggleton
  2011-08-26 16:18 ` Saul Wold
  0 siblings, 1 reply; 6+ messages in thread
From: Paul Eggleton @ 2011-08-26  9:47 UTC (permalink / raw)
  To: openembedded-core

Hi all,

Seems we have a few overlapping images in OE-core, with not especially obvious 
naming. The following two seem very similar in scope (basic X11 image):

meta/recipes-core/images/core-image-core.bb
meta/recipes-extended/images/core-image-basic.bb

Maybe we could just have one of these and call it core-image-x11-base or 
similar?

Then, we have core-image-base, which whilst it doesn't remove package 
management files, does not have "package-management" in its features, so it's 
not a whole lot different to core-image-minimal AFAICT.

Comments?

Cheers,
Paul


-- 

Paul Eggleton
Intel Open Source Technology Centre



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Core image recipes
  2011-08-26  9:47 Core image recipes Paul Eggleton
@ 2011-08-26 16:18 ` Saul Wold
  2011-08-26 16:39   ` Paul Eggleton
  0 siblings, 1 reply; 6+ messages in thread
From: Saul Wold @ 2011-08-26 16:18 UTC (permalink / raw)
  To: Patches and discussions about the oe-core layer; +Cc: Paul Eggleton

On 08/26/2011 02:47 AM, Paul Eggleton wrote:
> Hi all,
>
> Seems we have a few overlapping images in OE-core, with not especially obvious
> naming. The following two seem very similar in scope (basic X11 image):
>
> meta/recipes-core/images/core-image-core.bb

This contains X11_IMAGE_FEATURES

> meta/recipes-extended/images/core-image-basic.bb
>
This image should NOT contain any X11, this is supposed to be an 
extention of core-image-minimal with many of the busybox related 
commands substituted for the real command set.  The intention of this 
image is two fold, first it's the largest image that we test against 
non-GPLv3 and it's the non-graphical LSB image (I am not sure if there 
is a spec test defined for that.

> Maybe we could just have one of these and call it core-image-x11-base or
> similar?
>
So, no I do not think they can be merged.  But renaming core-image-core 
to core-image-x11-base might make sense, also renaming core-image-basic 
to core-image-lsb-basic may clear things up.

> Then, we have core-image-base, which whilst it doesn't remove package
> management files, does not have "package-management" in its features, so it's
> not a whole lot different to core-image-minimal AFAICT.
>
On this one I might agree, I know that we have not built that image, nor 
does it seem to be used by anything else.

Sau!

> Comments?
>
> Cheers,
> Paul
>
>



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Core image recipes
  2011-08-26 16:18 ` Saul Wold
@ 2011-08-26 16:39   ` Paul Eggleton
  2011-09-07 13:18     ` Paul Eggleton
  0 siblings, 1 reply; 6+ messages in thread
From: Paul Eggleton @ 2011-08-26 16:39 UTC (permalink / raw)
  To: Saul Wold; +Cc: Patches and discussions about the oe-core layer

On Friday 26 August 2011 17:18:15 Saul Wold wrote:
> On 08/26/2011 02:47 AM, Paul Eggleton wrote:
> > meta/recipes-extended/images/core-image-basic.bb
> 
> This image should NOT contain any X11, this is supposed to be an
> extention of core-image-minimal with many of the busybox related
> commands substituted for the real command set.  The intention of this
> image is two fold, first it's the largest image that we test against
> non-GPLv3 and it's the non-graphical LSB image (I am not sure if there
> is a spec test defined for that.

Hmm, I'm not sure what I was thinking earlier, you're right it doesn't appear 
to have X. I can't access the LSB specs website right now unfortunately but 
does this have an official name within LSB? It's not "LSB-Core" is it?
 
> So, no I do not think they can be merged.  But renaming core-image-core
> to core-image-x11-base might make sense, also renaming core-image-basic
> to core-image-lsb-basic may clear things up.

Sounds good to me. Should I send a patch?

> > Then, we have core-image-base, which whilst it doesn't remove package
> > management files, does not have "package-management" in its features, so
> > it's not a whole lot different to core-image-minimal AFAICT.
> 
> On this one I might agree, I know that we have not built that image, nor
> does it seem to be used by anything else.

If there's demand for a minimal image with package management (someone asked 
for this on IRC just the other day, and it makes sense to me at least) then 
that's what I'd suggest turning this into. In which case it ought to be called 
core-image-minimal-pkgmgmt or something similar.

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Core image recipes
  2011-08-26 16:39   ` Paul Eggleton
@ 2011-09-07 13:18     ` Paul Eggleton
  2011-09-08  3:53       ` Saul Wold
  0 siblings, 1 reply; 6+ messages in thread
From: Paul Eggleton @ 2011-09-07 13:18 UTC (permalink / raw)
  To: Patches and discussions about the oe-core layer

On Friday 26 August 2011 17:39:09 Paul Eggleton wrote:
> On Friday 26 August 2011 17:18:15 Saul Wold wrote:
> > On 08/26/2011 02:47 AM, Paul Eggleton wrote:
> > > meta/recipes-extended/images/core-image-basic.bb
> > 
> > This image should NOT contain any X11, this is supposed to be an
> > extention of core-image-minimal with many of the busybox related
> > commands substituted for the real command set.  The intention of this
> > image is two fold, first it's the largest image that we test against
> > non-GPLv3 and it's the non-graphical LSB image (I am not sure if there
> > is a spec test defined for that.
> 
> Hmm, I'm not sure what I was thinking earlier, you're right it doesn't
> appear to have X. 

Now I know why I thought this. Because task-core.bb defines task packages that 
depend on X applications, any recipe that inherits from core-image will force 
a build of all of the X apps even if it doesn't intend to use them - so both 
core-image-basic and core-image-base suffer from this.  This is not really very 
good and I think we ought to be splitting up task-core to avoid this. FYI 
whilst core-image-minimal inherits from core-image it overrides IMAGE_INSTALL 
and thus doesn't use anything from task-core and therefore does not have this 
issue.

> I can't access the LSB specs website right now
> unfortunately but does this have an official name within LSB? It's not
> "LSB-Core" is it?

Yep, it's LSB-Core (yet another meaning of "core", sigh...)
 
> > > Then, we have core-image-base, which whilst it doesn't remove package
> > > management files, does not have "package-management" in its features,
> > > so it's not a whole lot different to core-image-minimal AFAICT.
> > 
> > On this one I might agree, I know that we have not built that image, nor
> > does it seem to be used by anything else.
> 
> If there's demand for a minimal image with package management (someone
> asked for this on IRC just the other day, and it makes sense to me at
> least) then that's what I'd suggest turning this into. In which case it
> ought to be called core-image-minimal-pkgmgmt or something similar.

Any opinions on this one?

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Core image recipes
  2011-09-07 13:18     ` Paul Eggleton
@ 2011-09-08  3:53       ` Saul Wold
  2011-09-08  7:35         ` Paul Eggleton
  0 siblings, 1 reply; 6+ messages in thread
From: Saul Wold @ 2011-09-08  3:53 UTC (permalink / raw)
  To: Paul Eggleton; +Cc: Patches and discussions about the oe-core layer

On 09/07/2011 06:18 AM, Paul Eggleton wrote:
> On Friday 26 August 2011 17:39:09 Paul Eggleton wrote:
>> On Friday 26 August 2011 17:18:15 Saul Wold wrote:
>>> On 08/26/2011 02:47 AM, Paul Eggleton wrote:
>>>> meta/recipes-extended/images/core-image-basic.bb
>>>
>>> This image should NOT contain any X11, this is supposed to be an
>>> extention of core-image-minimal with many of the busybox related
>>> commands substituted for the real command set.  The intention of this
>>> image is two fold, first it's the largest image that we test against
>>> non-GPLv3 and it's the non-graphical LSB image (I am not sure if there
>>> is a spec test defined for that.
>>
>> Hmm, I'm not sure what I was thinking earlier, you're right it doesn't
>> appear to have X.
>
> Now I know why I thought this. Because task-core.bb defines task packages that
> depend on X applications, any recipe that inherits from core-image will force
> a build of all of the X apps even if it doesn't intend to use them - so both
> core-image-basic and core-image-base suffer from this.  This is not really very
> good and I think we ought to be splitting up task-core to avoid this. FYI
> whilst core-image-minimal inherits from core-image it overrides IMAGE_INSTALL
> and thus doesn't use anything from task-core and therefore does not have this
> issue.
>
Now I understand what you are talking about, it might best to split this 
into 2 tasks a task-core moved to recipes-core/tasks and a 
task-core-x11, what about that?

>> I can't access the LSB specs website right now
>> unfortunately but does this have an official name within LSB? It's not
>> "LSB-Core" is it?
>
> Yep, it's LSB-Core (yet another meaning of "core", sigh...)
>
We could rename basic to task-lsb-core if that's what your thinking, but 
as you point out yet another "core".


>>>> Then, we have core-image-base, which whilst it doesn't remove package
>>>> management files, does not have "package-management" in its features,
>>>> so it's not a whole lot different to core-image-minimal AFAICT.
>>>
>>> On this one I might agree, I know that we have not built that image, nor
>>> does it seem to be used by anything else.
>>
>> If there's demand for a minimal image with package management (someone
>> asked for this on IRC just the other day, and it makes sense to me at
>> least) then that's what I'd suggest turning this into. In which case it
>> ought to be called core-image-minimal-pkgmgmt or something similar.
>
> Any opinions on this one?
>
I think this is one that they can create themselves it's would be distro 
specific and would require additional space allocated to the rootfs, 
best for the distro do.  Remember we are trying to provide foundations 
and examples. core-image-minimal is supposed to be the smallest possible 
image with login and shell. It can be used by someone to build on.

Sau!


> Cheers,
> Paul
>



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Core image recipes
  2011-09-08  3:53       ` Saul Wold
@ 2011-09-08  7:35         ` Paul Eggleton
  0 siblings, 0 replies; 6+ messages in thread
From: Paul Eggleton @ 2011-09-08  7:35 UTC (permalink / raw)
  To: Saul Wold; +Cc: Patches and discussions about the oe-core layer

On Thursday 08 September 2011 04:53:28 you wrote:
> Now I understand what you are talking about, it might best to split this
> into 2 tasks a task-core moved to recipes-core/tasks and a
> task-core-x11, what about that?

I think that's the only way this can really work, yes.
 
> >> I can't access the LSB specs website right now
> >> unfortunately but does this have an official name within LSB? It's not
> >> "LSB-Core" is it?
> > 
> > Yep, it's LSB-Core (yet another meaning of "core", sigh...)
> 
> We could rename basic to task-lsb-core if that's what your thinking, but
> as you point out yet another "core".

Then again, if you're talking about LSB it makes sense to use the correct 
terminology.

> >> If there's demand for a minimal image with package management (someone
> >> asked for this on IRC just the other day, and it makes sense to me at
> >> least) then that's what I'd suggest turning this into. In which case it
> >> ought to be called core-image-minimal-pkgmgmt or something similar.
> > 
> > Any opinions on this one?
> 
> I think this is one that they can create themselves it's would be distro
> specific and would require additional space allocated to the rootfs,
> best for the distro do.  Remember we are trying to provide foundations
> and examples. core-image-minimal is supposed to be the smallest possible
> image with login and shell. It can be used by someone to build on.

That's exactly what I am thinking about. Since we already have such an image 
(core-image-base) that isn't being used for anything else, why not make it 
into something useful? Right now since core-image-minimal overrides 
IMAGE_INSTALL you can't use IMAGE_FEATURES and POKY_EXTRA_INSTALL to extend 
it, nor does it have package management that some users would be expecting; I 
think it would be useful to have a base image where those mechanisms do work. 
(Maybe the term "minimal" wouldn't apply to this image.)

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre



^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2011-09-08  7:40 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-08-26  9:47 Core image recipes Paul Eggleton
2011-08-26 16:18 ` Saul Wold
2011-08-26 16:39   ` Paul Eggleton
2011-09-07 13:18     ` Paul Eggleton
2011-09-08  3:53       ` Saul Wold
2011-09-08  7:35         ` Paul Eggleton

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.