ubuntu-phone team mailing list archive

[Michał Sawicz <michal.sawicz@xxxxxxxxxxxxx>]

W dniu 08.10.2015 o 12:48, Simon Fels pisze:
> On 08.10.2015 12:16, Michał Sawicz wrote:
>> W dniu 08.10.2015 o 11:54, Michael Zanetti pisze:
>>> Not sure if the Android layer is really the place to go. What happens
>>> when we run on an x86 based tablet which doesn't have the Android bits?
>>> Will that always have the DMI interface instead?
> 
> x86 based tablets should have the DMI interface.
> 
>> I agree, we need to plan for not having Android around. IMO we should
>> have a device file shipped, as you said, in the device tarball, that
>> configures the device. The user should be able to easily override things
>> stored there (for development purposes or other).
> 
> Sounds good and it looks like we already have a file for this:
> 
> $ cat /etc/ubuntu-touch-session.d/android.conf
> GRID_UNIT_PX=17
> QTWEBKIT_DPR=2.0
> 
> So what about just extending this one which comes either from the device
> tarball (krillin, arale) or from the ubuntu-touch-session package (flo,
> mako)?

Longer-term I think we should go away from putting it all in environment
(which is where they end up with) and use a richer file format (.ini,
.json, .yaml seems to be a choice often these days).

Until then, I'm fine with putting it there.

>> Priority should be:
>> - user override
>> - device tarball
>> - auto-detected fallback
> 
> Sure. Where a user override exists then this one should be applied
> first. In the case of the Bluetooth device class the user doesn't have
> any control over this.

Is there a reason why we would disallow overriding this? We'll have
devices that can't easily be categorized as one (I mean, is a 10" tablet
with GSM a phone or a tablet?).

-- 
Michał Sawicz <michal.sawicz@xxxxxxxxxxxxx>
Canonical Ltd.

Attachment: signature.asc
Description: OpenPGP digital signature