questions about bitbake world

questions about bitbake world

[Robert Yang]

Hello experts,

In the following code:

def buildWorldTargetList(self):
...
         for f in self.status.possible_world:
             terminal = True
             pn = self.status.pkg_fn[f]

             for p in pn_provides[pn]:
                 if p.startswith('virtual/'):
                     parselog.debug(2, "World build skipping %s due to %s 
provider starting with virtual/", f, p)
                     terminal = False
                     break
...

Does it mean that if a recipe contains:

PROVIDES = "virtual/something"

Then it would not be built until a pkg depends on it? Why we design it in this
way, please?

-- 
Thanks

Robert

Re: questions about bitbake world

[Chris Larson]

On Tue, Apr 10, 2012 at 1:24 AM, Robert Yang <liezhi.yang@windriver.com> wrote:
> In the following code:
>
> def buildWorldTargetList(self):
> ...
>        for f in self.status.possible_world:
>            terminal = True
>            pn = self.status.pkg_fn[f]
>
>            for p in pn_provides[pn]:
>                if p.startswith('virtual/'):
>                    parselog.debug(2, "World build skipping %s due to %s
> provider starting with virtual/", f, p)
>                    terminal = False
>                    break
> ...
>
> Does it mean that if a recipe contains:
>
> PROVIDES = "virtual/something"
>
> Then it would not be built until a pkg depends on it? Why we design it in
> this
> way, please?

The original intent with world was that when two recipes provided the
same thing, we'd still only want to build one of them, not both, as
building both can cause problems with stepping on one another in
sysroot, etc.
-- 
Christopher Larson

Re: questions about bitbake world

[Robert Yang]

On 04/10/2012 09:58 PM, Chris Larson wrote:
> On Tue, Apr 10, 2012 at 1:24 AM, Robert Yang<liezhi.yang@windriver.com>  wrote:
>> In the following code:
>>
>> def buildWorldTargetList(self):
>> ...
>>         for f in self.status.possible_world:
>>             terminal = True
>>             pn = self.status.pkg_fn[f]
>>
>>             for p in pn_provides[pn]:
>>                 if p.startswith('virtual/'):
>>                     parselog.debug(2, "World build skipping %s due to %s
>> provider starting with virtual/", f, p)
>>                     terminal = False
>>                     break
>> ...
>>
>> Does it mean that if a recipe contains:
>>
>> PROVIDES = "virtual/something"
>>
>> Then it would not be built until a pkg depends on it? Why we design it in
>> this
>> way, please?
>
> The original intent with world was that when two recipes provided the
> same thing, we'd still only want to build one of them, not both, as
> building both can cause problems with stepping on one another in
> sysroot, etc.

Thanks for your reply, but it seems that it builts none currently.

// Robert

Re: questions about bitbake world

[Chris Larson]

On Tue, Apr 10, 2012 at 7:12 AM, Robert Yang <liezhi.yang@windriver.com> wrote:
> On 04/10/2012 09:58 PM, Chris Larson wrote:
>>
>> On Tue, Apr 10, 2012 at 1:24 AM, Robert Yang<liezhi.yang@windriver.com>
>>  wrote:
>>>
>>> In the following code:
>>>
>>> def buildWorldTargetList(self):
>>> ...
>>>        for f in self.status.possible_world:
>>>            terminal = True
>>>            pn = self.status.pkg_fn[f]
>>>
>>>            for p in pn_provides[pn]:
>>>                if p.startswith('virtual/'):
>>>                    parselog.debug(2, "World build skipping %s due to %s
>>> provider starting with virtual/", f, p)
>>>                    terminal = False
>>>                    break
>>> ...
>>>
>>> Does it mean that if a recipe contains:
>>>
>>> PROVIDES = "virtual/something"
>>>
>>> Then it would not be built until a pkg depends on it? Why we design it in
>>> this
>>> way, please?
>>
>>
>> The original intent with world was that when two recipes provided the
>> same thing, we'd still only want to build one of them, not both, as
>> building both can cause problems with stepping on one another in
>> sysroot, etc.
>
>
> Thanks for your reply, but it seems that it builts none currently.

Fair point, I think in certain cases we also only wanted to build
things if they got pulled in via a dependency. For example, iirc
native and cross both exclude themselves from world via a variable set
in the classes. I don't recall if such was the case here, however.
-- 
Christopher Larson

Re: questions about bitbake world

[Robert Yang]

Thanks, is there a way to skip the pkg if I don't want to build it
when bitbake world, please?

// Robert

On 04/10/2012 10:15 PM, Chris Larson wrote:
> On Tue, Apr 10, 2012 at 7:12 AM, Robert Yang<liezhi.yang@windriver.com>  wrote:
>> On 04/10/2012 09:58 PM, Chris Larson wrote:
>>>
>>> On Tue, Apr 10, 2012 at 1:24 AM, Robert Yang<liezhi.yang@windriver.com>
>>>   wrote:
>>>>
>>>> In the following code:
>>>>
>>>> def buildWorldTargetList(self):
>>>> ...
>>>>         for f in self.status.possible_world:
>>>>             terminal = True
>>>>             pn = self.status.pkg_fn[f]
>>>>
>>>>             for p in pn_provides[pn]:
>>>>                 if p.startswith('virtual/'):
>>>>                     parselog.debug(2, "World build skipping %s due to %s
>>>> provider starting with virtual/", f, p)
>>>>                     terminal = False
>>>>                     break
>>>> ...
>>>>
>>>> Does it mean that if a recipe contains:
>>>>
>>>> PROVIDES = "virtual/something"
>>>>
>>>> Then it would not be built until a pkg depends on it? Why we design it in
>>>> this
>>>> way, please?
>>>
>>>
>>> The original intent with world was that when two recipes provided the
>>> same thing, we'd still only want to build one of them, not both, as
>>> building both can cause problems with stepping on one another in
>>> sysroot, etc.
>>
>>
>> Thanks for your reply, but it seems that it builts none currently.
>
> Fair point, I think in certain cases we also only wanted to build
> things if they got pulled in via a dependency. For example, iirc
> native and cross both exclude themselves from world via a variable set
> in the classes. I don't recall if such was the case here, however.

Re: questions about bitbake world

[Christopher Larson]

[-- Attachment #1: Type: text/plain, Size: 2093 bytes --]

Define EXCLUDE_FROM_WORLD = "1" in the recipe.

-- 
Christopher Larson


On Thursday, April 12, 2012 at 5:52 PM, Robert Yang wrote:

> 
> Thanks, is there a way to skip the pkg if I don't want to build it
> when bitbake world, please?
> 
> // Robert
> 
> On 04/10/2012 10:15 PM, Chris Larson wrote:
> > On Tue, Apr 10, 2012 at 7:12 AM, Robert Yang<liezhi.yang@windriver.com (mailto:liezhi.yang@windriver.com)> wrote:
> > > On 04/10/2012 09:58 PM, Chris Larson wrote:
> > > > 
> > > > On Tue, Apr 10, 2012 at 1:24 AM, Robert Yang<liezhi.yang@windriver.com (mailto:liezhi.yang@windriver.com)>
> > > > wrote:
> > > > > 
> > > > > In the following code:
> > > > > 
> > > > > def buildWorldTargetList(self):
> > > > > ...
> > > > > for f in self.status.possible_world:
> > > > > terminal = True
> > > > > pn = self.status.pkg_fn[f]
> > > > > 
> > > > > for p in pn_provides[pn]:
> > > > > if p.startswith('virtual/'):
> > > > > parselog.debug(2, "World build skipping %s due to %s
> > > > > provider starting with virtual/", f, p)
> > > > > terminal = False
> > > > > break
> > > > > ...
> > > > > 
> > > > > Does it mean that if a recipe contains:
> > > > > 
> > > > > PROVIDES = "virtual/something"
> > > > > 
> > > > > Then it would not be built until a pkg depends on it? Why we design it in
> > > > > this
> > > > > way, please?
> > > > > 
> > > > 
> > > > 
> > > > 
> > > > The original intent with world was that when two recipes provided the
> > > > same thing, we'd still only want to build one of them, not both, as
> > > > building both can cause problems with stepping on one another in
> > > > sysroot, etc.
> > > > 
> > > 
> > > 
> > > 
> > > Thanks for your reply, but it seems that it builts none currently.
> > 
> > Fair point, I think in certain cases we also only wanted to build
> > things if they got pulled in via a dependency. For example, iirc
> > native and cross both exclude themselves from world via a variable set
> > in the classes. I don't recall if such was the case here, however.
> > 
> 
> 
> 



[-- Attachment #2: Type: text/html, Size: 3125 bytes --]