FW: xrandr and xwayland

[David Deyo]

Thanks Daniel,

My local test is just a timer that rotates every 5 seconds.  I have a python keyboard running that rotates, but I have remnants of the original keyboard.
Almost like I have a stale buffer.  The extra keyboard images don’t respond either.

Is there some clear I need to be calling before I rotate?

-dwd

                                                                         David Deyo
[cid:image003.png at 01D78AB0.80A0FBB0]

Firmware Engineer
TPI- Tire Profiles
O: 214-396-3063
E:   ddeyo at tireprofiles.com<mailto:ddeyo at tireprofiles.com> | W:  www.tireprofiles.com<http://www.tireprofiles.com/>
A: 3010 Story Rd W, Irving, TX 75038


From: Daniel Stone<mailto:daniel at fooishbar.org>
Sent: Friday, August 6, 2021 3:14 AM
To: David Deyo<mailto:ddeyo at tireprofiles.com>
Cc: Emmanuel Gil Peyrot<mailto:linkmauve at linkmauve.fr>; wayland-devel at lists.freedesktop.org<mailto:wayland-devel at lists.freedesktop.org>
Subject: Re: FW: xrandr and xwayland

CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.


Hi David,

On Thu, 5 Aug 2021 at 22:17, David Deyo <ddeyo at tireprofiles.com> wrote:
> > Sounds like you're missing wl_display_flush() in your client code, so the requests don't make it to the socket buffer until they're forced to because it's filled up.
>
> That did it.  You guys are awesome.  I don’t suppose there’s a Weston doc somewhere that would have told me that, had I looked.

It's a little bit buried, but this is the best explanation of how to
integrate Wayland into an event loop, as you would with a toolkit:
  https://wayland.freedesktop.org/docs/html/apb.html#Client-classwl__display_1a40039c1169b153269a3dc0796a54ddb0

If you scroll up to the main wl_display section, it explains how event
queues are used as well.

Broadly speaking, the advice is to:
- when active, process your events and send any requests
- immediately before you go into a passive state (waiting for events,
sleeping, etc), flush the display so your requests get delivered
- run the prepare_read_queue() / dispatch_queue_pending() loop
immediately before sleeping, in order to make sure you get all events
queued for you, then flush again in case you've queued any requests
from your event handlers
- poll on the Wayland display FD as well as any other activity sources
(other event queues, timers, etc)
- when you wake up, dispatch your Wayland event queue as well as other
relevant event sources

> > Also, my taskbar is the wrong length and my background is black.  Other than that, pretty cool.
>
> Yep, desktop-shell isn't designed to handle runtime rotation. It could be made to pretty easily by working on the client code. For your case though I'd assume something like kiosk-shell would be a much better bet.

kiosk-shell is something we have in newer versions of Weston which
sounds like it would work well for your usecases - it's designed to
just run a single application fullscreen. You might want to check out
what we have in git, which will be released as 10.0 in a few weeks'
time.

The rotation patches never got merged because we had some issues with
the IIO integration in particular, but having runtime rotation tests
sure would be nice, and kiosk-shell should at least be a lot easier to
fix than desktop-shell, if it does even need any fixes.

Cheers,
Daniel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/wayland-devel/attachments/20210806/ef65abff/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 5A8379ED5FC7400BA9F6BAB3032E5B6D.png
Type: image/png
Size: 10540 bytes
Desc: 5A8379ED5FC7400BA9F6BAB3032E5B6D.png
URL: <https://lists.freedesktop.org/archives/wayland-devel/attachments/20210806/ef65abff/attachment-0001.png>