[PATCH weston 00/14] Desktop Protocol Support for IVI-Shell

[Matt Hoosier]

On Tue, Nov 7, 2017 at 11:58 AM, Quentin Glidic
<sardemff7+wayland at sardemff7.net> wrote:
> On 11/7/17 6:01 PM, Matt Hoosier wrote:
>>
>> Hi Pekka,
>>
>> On Wed, Oct 25, 2017 at 10:09 AM, Ucan, Emre (ADITG/ESB)
>> <eucan at de.adit-jv.com> wrote:
>>>
>>> Actually,  IMO ivi-shell is not a proper wayland compositor.
>>> Because it is violating wayland protocol by not supporting wl_shell
>>> interface.
>
>
> The wl_shell protocol was designed for the desktop use case. So not
> supporting it is perfectly fine on non-desktop.
>
>
>>> Therefore, we have to at least support wl_shell interface in
>>> ivi-shell. Why not support it via libweston-desktop ?
>>
>>
>> I'm wondering if you have any thoughts on this one specific point
>> that Emre made. I know there's a lot of heartburn over the inclusion
>> of wl_shell into the core protocol, and you wouldn't do it that way
>> if that decision were getting made today.
>>
>> But given the history that actually happened, is there a reason not
>> to go ahead and allow the ivi-shell to implement wl_shell simply on
>> the grounds that it is part of the defined core protocol? I think
>> that some potentially reasonable answers were made above to your
>> concerns that the API offered by wl_shell targeted toward desktops
>> wouldn't be meaningful on an IVI system.  All the mandatory
>> operations seem to be possible to support,
>
>
> Except the current patch doesn’t say that.

I'm unclear whether you're objecting to the verbage in the changeset
frontmatter. Are you just calling for Michael to state this ground in
the commit message?

> xdg_shell allows the compositor
> to ignore the fullscreen/maximize requests. libweston-desktop API was
> designed around xdg_shell, with best effort for wl_shell (and I am in the
> process of fixing that).

I'm trying to figure out whether Pekka is unconditionally opposed to
adding any desktop-ish API support to the IVI compositor. For the
moment (see my "magic wand" comment earlier), I think it's useful to
suppose that using libweston did not commit the shell to supporting
both xdg and wl_shell as a packaged deal. If there's still no consent
toward this patch series even with that liberating assumption, then no
amount of implementation adjustment would probably be found
persuasive.

> With wl_shell, the compositor cannot deny the
> client the fullscreen or maximized state.

Are you pointing to a behavior in Michael's proposed implementation
that dishonors a request from a wl_shell client to enter fullscreen or
maximized state? If that's the case, maybe he's willing to make a
guarantee in the IVI shell a weston_desktop_surface always gets such
states honored.

>
>
>> and the IVI shell just needs to come up with sensible definitions
>> (i.e., documented for its users) about how the anonymous wl_shell
>> clients' surfaces will be integrated with the explicit IVI clients.
>>
>> Note, I'm not for the moment trying to expand this line of reasoning into
>> a grounds for justifying the support of xdg-shell. That
>> protocol is (deliberately) not part of the core, and I understand
>> that. If a magic wand were to be waved and use of libweston-desktop
>> didn't automatically mean that xdg-shell is supported too, would that
>> be tolerable?
>
>
> xdg_shell was not added to core to avoid the wl_shell situation. We cannot
> drop wl_shell support because it’s in core wayland.xml, and it’s hurting us
> already, because wl_shell clients will always prevent a full migration to
> xdg_shell. We are lucky that libweston-desktop (and others) doesn’t need too
> much code to support it.
>
>
> Thanks,
>
> --
>
> Quentin “Sardem FF7” Glidic