Where should MACHINE_TASK_PROVIDER be defined?

Where should MACHINE_TASK_PROVIDER be defined?

[Rod Whitby]

There seems to be significant inconsistency in the way that
MACHINE_TASK_PROVIDER is defined in OE (see the output of a recursive
grep command in .../conf at the end of this message).

It's defaulted to "task-base" in bitbake.conf, and then every man and
his dog forcibly defines it to the same thing in a variety of
machine.conf files and distro.conf files.

Should all of these instances simply be deleted, cause none of them
actually change the default, and all of them stop the user from changing
 it in local.conf ...

If it is to be overridden, should it be a distro override or a machine
override?  What would be normal reasons for overriding it?

> find . -type f -print0 | xargs -0 grep MACHINE_TASK_PROVIDER
> ./distro/include/angstrom.inc:MACHINE_TASK_PROVIDER = "task-base"
> ./distro/include/openmoko.inc:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/collie.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/h3600.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/jornada6xx.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/jornada7xx.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/mainstone.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/netvista.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/nokia770.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/omap5912osk.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/include/zaurus-2.6.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/include/lsppc.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/simpad.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/x86.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/devkitidp-pxa255.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/qemuarm.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/efika.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/logicpd-pxa270.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/htcuniversal.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/dht-walnut.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/titan.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/wrap.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/smdk2440.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/compulab-pxa270.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/storcenter.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/qemux86.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/fic-gta01.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/shark.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/turbostation.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/n2100.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/smdk2443.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./machine/lsarm.conf:MACHINE_TASK_PROVIDER = "task-base"
> ./bitbake.conf:MACHINE_TASK_PROVIDER ?= "task-base"

-- Rod

Re: Where should MACHINE_TASK_PROVIDER be defined?

[Marcin Juszkiewicz]

Dnia piątek, 13 kwietnia 2007, Rod Whitby napisał:
> There seems to be significant inconsistency in the way that
> MACHINE_TASK_PROVIDER is defined in OE (see the output of a recursive
> grep command in .../conf at the end of this message).
>
> It's defaulted to "task-base" in bitbake.conf, and then every man and
> his dog forcibly defines it to the same thing in a variety of
> machine.conf files and distro.conf files.
>
> Should all of these instances simply be deleted

Yes - most of them are from time when we had task-bootstrap as default.

-- 
JID: hrw-jabber.org
OpenEmbedded developer/consultant

         Packrat's Credo:
         	I have no use for it, but I hate to see it go to waste.

Re: Where should MACHINE_TASK_PROVIDER be defined?

[Richard Purdie]

On Fri, 2007-04-13 at 11:48 +0200, Marcin Juszkiewicz wrote:
> Dnia piątek, 13 kwietnia 2007, Rod Whitby napisał:
> > There seems to be significant inconsistency in the way that
> > MACHINE_TASK_PROVIDER is defined in OE (see the output of a recursive
> > grep command in .../conf at the end of this message).
> >
> > It's defaulted to "task-base" in bitbake.conf, and then every man and
> > his dog forcibly defines it to the same thing in a variety of
> > machine.conf files and distro.conf files.
> >
> > Should all of these instances simply be deleted
> 
> Yes - most of them are from time when we had task-bootstrap as default.

Agreed.

They're from the transition period when task-bootstrap was default and
each machine had to be tagged as it was converted. All the duplicates
can now be removed since the default has changed.

Cheers,

Richard 

Re: Where should MACHINE_TASK_PROVIDER be defined?

[Rod Whitby]

Richard Purdie wrote:
> On Fri, 2007-04-13 at 11:48 +0200, Marcin Juszkiewicz wrote:
>> Dnia piątek, 13 kwietnia 2007, Rod Whitby napisał:
>>> There seems to be significant inconsistency in the way that
>>> MACHINE_TASK_PROVIDER is defined in OE (see the output of a recursive
>>> grep command in .../conf at the end of this message).
>>>
>>> It's defaulted to "task-base" in bitbake.conf, and then every man and
>>> his dog forcibly defines it to the same thing in a variety of
>>> machine.conf files and distro.conf files.
>>>
>>> Should all of these instances simply be deleted
>> Yes - most of them are from time when we had task-bootstrap as default.
> 
> Agreed.
> 
> They're from the transition period when task-bootstrap was default and
> each machine had to be tagged as it was converted. All the duplicates
> can now be removed since the default has changed.

This has now been done and committed.

-- Rod

Re: Where should MACHINE_TASK_PROVIDER be defined?

[Michael 'Mickey' Lauer]

Rod Whitby wrote:
> This has now been done and committed.

Thanks for cleaning up.

Regards,

:M:
-- 
Michael 'Mickey' Lauer | IT-Freelancer | http://www.vanille-media.de