RDEPENDS_${PN} overwrites RDEPENDS_foo - bug or feature?

RDEPENDS_${PN} overwrites RDEPENDS_foo - bug or feature?

[Stanislav Brabec]

While verifying all aspects of proposed udev change, I experienced a
strange behavior of an udev recipe with RDEPENDS_udev:

RDEPENDS_udev is always ignored. Searching in deep, I found that
RDEPENDS_${PN} defined in update-rc.d.bbclass overwrites RDEPENDS_udev.

Once RDEPENDS_${PN} is defined, RDEPENDS_udev has no effect.

Is it a bug or documented feature?

-- 

________________________________________________________________________
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: RDEPENDS_${PN} overwrites RDEPENDS_foo - bug or feature?

[Frans Meulenbroeks]

2010/7/20 Stanislav Brabec <utx@penguin.cz>

> While verifying all aspects of proposed udev change, I experienced a
> strange behavior of an udev recipe with RDEPENDS_udev:
>
> RDEPENDS_udev is always ignored. Searching in deep, I found that
> RDEPENDS_${PN} defined in update-rc.d.bbclass overwrites RDEPENDS_udev.
>
> Once RDEPENDS_${PN} is defined, RDEPENDS_udev has no effect.
>

Not an expert in this, but would  RDEPENDS_${PN}_udev (or perhaps
RDEPENDS_udev_${PN}) do the trick ?
Ofc this is just a suggested workaround, no idea if what you describe is
desired or not.

Frans

>
> Is it a bug or documented feature?
>
> --
>
> ________________________________________________________________________
> Stanislav Brabec
> http://www.penguin.cz/~utx/zaurus <http://www.penguin.cz/%7Eutx/zaurus>
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: RDEPENDS_${PN} overwrites RDEPENDS_foo - bug or feature?

[Chris Larson]

On Tue, Jul 20, 2010 at 1:04 PM, Stanislav Brabec <utx@penguin.cz> wrote:

> While verifying all aspects of proposed udev change, I experienced a
> strange behavior of an udev recipe with RDEPENDS_udev:
>
> RDEPENDS_udev is always ignored. Searching in deep, I found that
> RDEPENDS_${PN} defined in update-rc.d.bbclass overwrites RDEPENDS_udev.
>
> Once RDEPENDS_${PN} is defined, RDEPENDS_udev has no effect.
>
> Is it a bug or documented feature?


Neither, it's just the way bitbake works today.  When the name of a variable
includes references to other variables like this, they aren't expanded until
the end of the parsing, and if a variable already existed with the new name,
it gets overwritten.  One could argue that bitbake should error in this
case, but that wouldn't happen anytime soon due to the change in behavior.
 It wouldn't go into 1.10, but whatever comes after that, and 1.10 isn't
even out yet.

In the meantime, you can check for this.  INHERIT += "recipe_sanity", then
bitbake -c recipe_sanity udev, or bitbake -c recipe_sanity_all udev to check
it and its dependencies.
-- 
Christopher Larson
clarson at kergoth dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Senior Software Engineer, Mentor Graphics

Re: RDEPENDS_${PN} overwrites RDEPENDS_foo - bug or feature?

[Phil Blundell]

On Tue, 2010-07-20 at 22:04 +0200, Stanislav Brabec wrote:
> While verifying all aspects of proposed udev change, I experienced a
> strange behavior of an udev recipe with RDEPENDS_udev:
> 
> RDEPENDS_udev is always ignored. Searching in deep, I found that
> RDEPENDS_${PN} defined in update-rc.d.bbclass overwrites RDEPENDS_udev.
> 
> Once RDEPENDS_${PN} is defined, RDEPENDS_udev has no effect.
> 
> Is it a bug or documented feature?

Possibly neither of the above.  It's not a bug, but I'm not certain that
it is very well documented either.

p.