Merge lp://qastaging/~gerboland/webbrowser-app/formFactor-support into lp://qastaging/webbrowser-app

I’m not seeing any formFactor property defined on UserAgent02, so I doubt the onFormFactorChanged handler in UbuntuWebContext.qml would work.

Note that I’m currently working on a branch that removes the 'formFactor' property from the Ubuntu.Web plugin’s context object¹, so it’s kind of ironical that you’re adding it back in another branch. I really don’t think the default user agent (and corresponding overrides) should be tied to the "form factor" (whatever that means in a world of convergence). How does mir determine whether a device is a phone or a desktop, or a tablet? Is that a solid API that we are going to support and publicize?

Even if that was a sensible thing to do, we most probably don’t want to get a mobile UX on a tablet. Well, at least not a 10" tablet. On a 7" tablet maybe. See where the screenDiagonal property comes from?

As a side note, the dependency on Qt5Gui_PRIVATE is not very welcome, given that we’ve been steadily working towards entirely removing the dependencies on private Qt packages (ask Mirv about it).

¹ https://code.launchpad.net/~osomon/webbrowser-app/remove-formFactor/+merge/285539

FAILED: Continuous integration, rev:1358
http://jenkins.qa.ubuntu.com/job/webbrowser-app-ci/2626/
Executed test runs:
    FAILURE: http://jenkins.qa.ubuntu.com/job/generic-deb-autopilot-vivid-touch/6587/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/webbrowser-app-vivid-amd64-ci/1379/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/webbrowser-app-vivid-armhf-ci/1379/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/webbrowser-app-vivid-i386-ci/1379/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/generic-mediumtests-builder-vivid-armhf/6598/console

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/webbrowser-app-ci/2626/rebuild

> I’m not seeing any formFactor property defined on UserAgent02, so I doubt the
> onFormFactorChanged handler in UbuntuWebContext.qml would work.

Code untested. Was struggling to cross compile & deploy, sbuild unhappy with me today.

> Note that I’m currently working on a branch that removes the 'formFactor'
> property from the Ubuntu.Web plugin’s context object¹, so it’s kind of
> ironical that you’re adding it back in another branch. I really don’t think
> the default user agent (and corresponding overrides) should be tied to the
> "form factor" (whatever that means in a world of convergence). How does mir
> determine whether a device is a phone or a desktop, or a tablet? Is that a
> solid API that we are going to support and publicize?

I obviously have no idea of your plans, this was just an attempt at replacing your screen heuristic with something more concrete.

Form Factor will be a hint to the application to tell it what mode the shell is in, on each screen. Mir supports it, shell will do its best to set it correctly. I would hope the average app author has no need for such a property, any form factor specific behaviour would be encapsulated in the UITK.

> Even if that was a sensible thing to do, we most probably don’t want to get a
> mobile UX on a tablet. Well, at least not a 10" tablet. On a 7" tablet maybe.
> See where the screenDiagonal property comes from?

Is your call. The goal of this was to indicate that you can learn if the app is running on a tablet or a desktop UI at the time.

> As a side note, the dependency on Qt5Gui_PRIVATE is not very welcome, given
> that we’ve been steadily working towards entirely removing the dependencies on
> private Qt packages (ask Mirv about it).

Qt doesn't have a nice way to export it to apps yet, hence having to resort to private Qt headers. As I mentioned in a comment, a better solution is for the UITK to do that for you. This isn't the final solution.
Thanks for the comments!

> > Even if that was a sensible thing to do, we most probably don’t want
> > to get a mobile UX on a tablet. Well, at least not a 10" tablet. On
> > a 7" tablet maybe. See where the screenDiagonal property comes from?
>
> Is your call. The goal of this was to indicate that you can learn if
> the app is running on a tablet or a desktop UI at the time.

Thanks, and that’s useful, but I’m not convinced that (or at least that *alone*) is enough to decide what the default UA string should be. I have started a conversation with design about that, we need to come to a consensus. Until that happens, I’d rather keep the current implementation based on screen size (I’m not against refining it though).

> > I’m not seeing any formFactor property defined on UserAgent02, so I doubt
> the
> > onFormFactorChanged handler in UbuntuWebContext.qml would work.
>
> Code untested. Was struggling to cross compile & deploy, sbuild unhappy with
> me today.
>
> > Note that I’m currently working on a branch that removes the 'formFactor'
> > property from the Ubuntu.Web plugin’s context object¹, so it’s kind of
> > ironical that you’re adding it back in another branch. I really don’t think
> > the default user agent (and corresponding overrides) should be tied to the
> > "form factor" (whatever that means in a world of convergence). How does mir
> > determine whether a device is a phone or a desktop, or a tablet? Is that a
> > solid API that we are going to support and publicize?
>
> I obviously have no idea of your plans, this was just an attempt at replacing
> your screen heuristic with something more concrete.
>
> Form Factor will be a hint to the application to tell it what mode the shell
> is in, on each screen. Mir supports it, shell will do its best to set it
> correctly. I would hope the average app author has no need for such a
> property, any form factor specific behaviour would be encapsulated in the
> UITK.
>

It sounds like this might be useful for bug 1545088, bug 1547145 and bug 1547102. How does the shell determine what the hint is?

> > Even if that was a sensible thing to do, we most probably don’t want to get
> a
> > mobile UX on a tablet. Well, at least not a 10" tablet. On a 7" tablet
> maybe.
> > See where the screenDiagonal property comes from?
>
> Is your call. The goal of this was to indicate that you can learn if the app
> is running on a tablet or a desktop UI at the time.
>
> > As a side note, the dependency on Qt5Gui_PRIVATE is not very welcome, given
> > that we’ve been steadily working towards entirely removing the dependencies
> on
> > private Qt packages (ask Mirv about it).
>
> Qt doesn't have a nice way to export it to apps yet, hence having to resort to
> private Qt headers. As I mentioned in a comment, a better solution is for the
> UITK to do that for you. This isn't the final solution.
> Thanks for the comments!

FAILED: Continuous integration, rev:1359
http://jenkins.qa.ubuntu.com/job/webbrowser-app-ci/2627/
Executed test runs:
    UNSTABLE: http://jenkins.qa.ubuntu.com/job/generic-deb-autopilot-vivid-touch/6588
    SUCCESS: http://jenkins.qa.ubuntu.com/job/webbrowser-app-vivid-amd64-ci/1380
    SUCCESS: http://jenkins.qa.ubuntu.com/job/webbrowser-app-vivid-armhf-ci/1380
        deb: http://jenkins.qa.ubuntu.com/job/webbrowser-app-vivid-armhf-ci/1380/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/webbrowser-app-vivid-i386-ci/1380
    UNSTABLE: http://jenkins.qa.ubuntu.com/job/generic-deb-autopilot-runner-vivid-touch/5011
    SUCCESS: http://jenkins.qa.ubuntu.com/job/generic-mediumtests-builder-vivid-armhf/6599
        deb: http://jenkins.qa.ubuntu.com/job/generic-mediumtests-builder-vivid-armhf/6599/artifact/work/output/*zip*/output.zip
    SUCCESS: http://s-jenkins.ubuntu-ci:8080/job/touch-flash-device/27786

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/webbrowser-app-ci/2627/rebuild