Handling machines of the same architecture by different screen resolutions

Handling machines of the same architecture by different screen resolutions

[Paul Sokolovsky]

Hello openembedded-devel,

  Continuing discussion of how to handle such important machine
property like screen size/resolution, here's the issue: there're
cases, when same "machine" in OE terms actually has submodels which
differ in screen resolution. One example is recently added aximx50
machine. That machine covers Axim X50, Axim X50V, Axim X51, Axim X51V
PDA models. "V" models have VGA screen, others - QVGA. They are served
by the same kernel port, though defconfigs for "V" models differ, as
there's no runtime detection.

  And, again, screen sizes differ, so "V" machines should be
bigscreen, while "V-less" - smallscreen. Does this warrant adding
another machine to OE to cover "V" models? (It would be obviously
done via refactoring common parts to machine/include/ ).

-- 
Best regards,
 Paul                          mailto:pmiscml@gmail.com

Re: Handling machines of the same architecture by different screen resolutions

[Richard Purdie]

On Wed, 2007-03-07 at 17:31 +0200, Paul Sokolovsky wrote:
> Hello openembedded-devel,
> 
>   Continuing discussion of how to handle such important machine
> property like screen size/resolution, here's the issue: there're
> cases, when same "machine" in OE terms actually has submodels which
> differ in screen resolution. One example is recently added aximx50
> machine. That machine covers Axim X50, Axim X50V, Axim X51, Axim X51V
> PDA models. "V" models have VGA screen, others - QVGA. They are served
> by the same kernel port, though defconfigs for "V" models differ, as
> there's no runtime detection.
> 
>   And, again, screen sizes differ, so "V" machines should be
> bigscreen, while "V-less" - smallscreen. Does this warrant adding
> another machine to OE to cover "V" models? (It would be obviously
> done via refactoring common parts to machine/include/ ).

In my view yes, since they need different kernels.

If they used a common kernel, it would be a choice for the machine
maintainer where both large and small icons would be needed in the
image, selected at runtime in the common image case. A separate machine
is probably easier.

Cheers,

Richard