bblayer priorities ignored for bbappends?

bblayer priorities ignored for bbappends?

[Tobias Olausson]

[-- Attachment #1: Type: text/plain, Size: 953 bytes --]

Hi!

I just had a weird experience with yocto and I wanted to know if the
described behaviour is intentional.

I have a layer with priority 7 containing a recipe, foo.bb. In another
layer, with priority 8, there is a foo.bbappend. Finally, I just added
another layer, with priority 9, where foo.bb is redefined as a binary
recipe.

When I added the last layer, I expected that to take precedence over the
other two. However, the bbappend for foo is still applied (which in this
case changed the SRC_URI, making the whole binary recipe moot). Is this
really the way it's supposed to be? I would expect that lower layers cannot
influence layers with higher priority at all, but that does not seem to be
the case.

Cheers,
Tobias Olausson
M.Sc
Software Engineer

PELAGICORE | Experience Change
Ekelundsgatan 4, 6tr, SE-411 18 Gothenburg, Sweden
Mobile: +46(0)735-873444
E-Mail: tobias.olausson@pelagicore.com
IRC: wto @ FreeNode

[-- Attachment #2: Type: text/html, Size: 1437 bytes --]

Re: bblayer priorities ignored for bbappends?

[Paul Eggleton]

Hi Tobias,

On Wednesday 08 April 2015 08:54:20 Tobias Olausson wrote:
> I just had a weird experience with yocto and I wanted to know if the
> described behaviour is intentional.
> 
> I have a layer with priority 7 containing a recipe, foo.bb. In another
> layer, with priority 8, there is a foo.bbappend. Finally, I just added
> another layer, with priority 9, where foo.bb is redefined as a binary
> recipe.
> 
> When I added the last layer, I expected that to take precedence over the
> other two. However, the bbappend for foo is still applied (which in this
> case changed the SRC_URI, making the whole binary recipe moot). Is this
> really the way it's supposed to be? I would expect that lower layers cannot
> influence layers with higher priority at all, but that does not seem to be
> the case.

That is the case, yes - bbappends are always applied. As far as I am aware 
this is the first time this kind of situation has been asked about though so 
it's probably just that we didn't consider it before.

Presumably you could use a bbappend rather than a recipe in the last layer? 
You might need to stub out some of the functions of course, but it should be 
possible.

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre

Re: bblayer priorities ignored for bbappends?

[Nikolay Dimitrov]

Hi Paul, Tobias,

On 04/08/2015 11:55 AM, Paul Eggleton wrote:
> Hi Tobias,
>
> On Wednesday 08 April 2015 08:54:20 Tobias Olausson wrote:
>> I just had a weird experience with yocto and I wanted to know if the
>> described behaviour is intentional.
>>
>> I have a layer with priority 7 containing a recipe, foo.bb. In another
>> layer, with priority 8, there is a foo.bbappend. Finally, I just added
>> another layer, with priority 9, where foo.bb is redefined as a binary
>> recipe.
>>
>> When I added the last layer, I expected that to take precedence over the
>> other two. However, the bbappend for foo is still applied (which in this
>> case changed the SRC_URI, making the whole binary recipe moot). Is this
>> really the way it's supposed to be? I would expect that lower layers cannot
>> influence layers with higher priority at all, but that does not seem to be
>> the case.
>
> That is the case, yes - bbappends are always applied. As far as I am aware
> this is the first time this kind of situation has been asked about though so
> it's probably just that we didn't consider it before.
>
> Presumably you could use a bbappend rather than a recipe in the last layer?
> You might need to stub out some of the functions of course, but it should be
> possible.

Thanks for touching the important subject. Indeed, this is an issue as
sometimes the bbappend change is generic and can be reused by the
customer layer(s), and sometimes the change is not appropriate and
needs to be undone.

I would agree that both cases are legitimate and would look for ideas
how it can be resolved in a logical way.

Regards,
Nikolay

PS: I can give examples, but don't want to drift the discussion towards
my topics.

Re: bblayer priorities ignored for bbappends?

[Tobias Olausson]

[-- Attachment #1: Type: text/plain, Size: 1836 bytes --]

Hi Paul

On 8 April 2015 at 10:55, Paul Eggleton <paul.eggleton@linux.intel.com>
wrote:

> Hi Tobias,
>
> On Wednesday 08 April 2015 08:54:20 Tobias Olausson wrote:
> > I just had a weird experience with yocto and I wanted to know if the
> > described behaviour is intentional.
> >
> > I have a layer with priority 7 containing a recipe, foo.bb. In another
> > layer, with priority 8, there is a foo.bbappend. Finally, I just added
> > another layer, with priority 9, where foo.bb is redefined as a binary
> > recipe.
> >
> > When I added the last layer, I expected that to take precedence over the
> > other two. However, the bbappend for foo is still applied (which in this
> > case changed the SRC_URI, making the whole binary recipe moot). Is this
> > really the way it's supposed to be? I would expect that lower layers
> cannot
> > influence layers with higher priority at all, but that does not seem to
> be
> > the case.
>
> That is the case, yes - bbappends are always applied. As far as I am aware
> this is the first time this kind of situation has been asked about though
> so
> it's probably just that we didn't consider it before.
>

I see. I can see a point of always applying bbappends, of course, I had
just never considered it either.


>
> Presumably you could use a bbappend rather than a recipe in the last layer?
> You might need to stub out some of the functions of course, but it should
> be
> possible.
>

Indeed, that is a reasonable solution.


>
> Cheers,
> Paul
>
> --
>
> Paul Eggleton
> Intel Open Source Technology Centre
>



-- 
Tobias Olausson
M.Sc
Software Engineer

PELAGICORE | Experience Change
Ekelundsgatan 4, 6tr, SE-411 18 Gothenburg, Sweden
Mobile: +46(0)735-873444
E-Mail: tobias.olausson@pelagicore.com
IRC: wto @ FreeNode

[-- Attachment #2: Type: text/html, Size: 3007 bytes --]

Re: bblayer priorities ignored for bbappends?

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 2573 bytes --]

On Wed, Apr 08, 2015 at 12:29:21PM +0200, Tobias Olausson wrote:
> Hi Paul
> 
> On 8 April 2015 at 10:55, Paul Eggleton <paul.eggleton@linux.intel.com>
> wrote:
> 
> > Hi Tobias,
> >
> > On Wednesday 08 April 2015 08:54:20 Tobias Olausson wrote:
> > > I just had a weird experience with yocto and I wanted to know if the
> > > described behaviour is intentional.
> > >
> > > I have a layer with priority 7 containing a recipe, foo.bb. In another
> > > layer, with priority 8, there is a foo.bbappend. Finally, I just added
> > > another layer, with priority 9, where foo.bb is redefined as a binary
> > > recipe.
> > >
> > > When I added the last layer, I expected that to take precedence over the
> > > other two. However, the bbappend for foo is still applied (which in this
> > > case changed the SRC_URI, making the whole binary recipe moot). Is this
> > > really the way it's supposed to be? I would expect that lower layers
> > cannot
> > > influence layers with higher priority at all, but that does not seem to
> > be
> > > the case.
> >
> > That is the case, yes - bbappends are always applied. As far as I am aware
> > this is the first time this kind of situation has been asked about though
> > so
> > it's probably just that we didn't consider it before.
> >
> 
> I see. I can see a point of always applying bbappends, of course, I had
> just never considered it either.
> 
> 
> >
> > Presumably you could use a bbappend rather than a recipe in the last layer?
> > You might need to stub out some of the functions of course, but it should
> > be
> > possible.
> >
> 
> Indeed, that is a reasonable solution.

It's also possible to use BBMASK to mask bbappends from lower layers in
cases where bbappends are hard/impossible to undo and you cannot easily
convert the recipe in top layer to bbappend (e.g. you don't want to,
because it's different version and lower bbappends are version agnostic
with %).

Regards,

> 
> 
> >
> > Cheers,
> > Paul
> >
> > --
> >
> > Paul Eggleton
> > Intel Open Source Technology Centre
> >
> 
> 
> 
> -- 
> Tobias Olausson
> M.Sc
> Software Engineer
> 
> PELAGICORE | Experience Change
> Ekelundsgatan 4, 6tr, SE-411 18 Gothenburg, Sweden
> Mobile: +46(0)735-873444
> E-Mail: tobias.olausson@pelagicore.com
> IRC: wto @ FreeNode

> -- 
> _______________________________________________
> yocto mailing list
> yocto@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/yocto


-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 188 bytes --]