From kde-panel-devel  Sun Apr 29 21:57:18 2018
From: David Edmundson <david () davidedmundson ! co ! uk>
Date: Sun, 29 Apr 2018 21:57:18 +0000
To: kde-panel-devel
Subject: Notes from Sprint - Wayland stuff
Message-Id: <CAGeFrHBWLC8PLgTpNxC9oqjWTPK0Y=4-ENjSfVJMqY7i94-yKA () mail ! gmail ! com>
X-MARC-Message: https://marc.info/?l=kde-panel-devel&m=152503907222319
MIME-Version: 1
Content-Type: multipart/mixed; boundary="--0000000000005bf4a9056b03d3d5"

--0000000000005bf4a9056b03d3d5
Content-Type: text/plain; charset="UTF-8"

I'm posting my notes from the sprint grouped by topic for easier tracking
of replies.

As you may know, we had the lead dev of wlroots at the sprint, with one
goal being to talk over custom protocols we both have.

I've attached my notes (and my opinions) below.

---

Layer protocol.

Effectively its a shell protocol with semantic xdgpositioner like
properties aligned to a wl_output rather than ours which uses global pixel
coordinates.

Its mostly quite good, I posted the few things I don't like to their
tracker. It would cover almost every plasmashell protocol usage except
panel autohide which is quite kwin specific.

Given our protocol works, I don't think we gain anything switching as is,
but if it was both standardised *and* if we can get it into Qt (properly
with abi stable api) I would more than happily add support.

-----
Output management protocol

Sway decided their users will just edit config files and reload compositor;
but did say they'd consider our protocol if they add it.

------

Remote access protocol.

wlroots has a draft DMA buf sharing protocol, which is conceptually similar
to our gem sharing but their protocol is much more...involved. It does a
lot more but at the expense of being quite complex for both the compositor
and any potential recording client.

Does show plenty we could add to ours, but we should see how stuff with
pipewire pans out first and what's actually needed.

David

--0000000000005bf4a9056b03d3d5
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>I&#39;m posting my notes from the sprint grouped by t=
opic for easier tracking of replies.<br></div><div><br>As you may know, we =
had the lead dev of wlroots at the sprint, with one goal being to talk over=
 custom protocols we both have.<br><br></div>I&#39;ve attached my notes (an=
d my opinions) below.<br><div><div><br>---<br><br>Layer protocol. <br><br>E=
ffectively its a shell protocol with semantic xdgpositioner like properties=
 aligned to a wl_output rather than ours which uses global pixel coordinate=
s.<br><br>Its mostly quite good, I posted the few things I don&#39;t like t=
o their tracker. It would cover almost every plasmashell protocol usage exc=
ept panel autohide which is quite kwin specific. <br><br>Given our protocol=
 works, I don&#39;t think we gain anything switching as  is, but if it was =
both standardised *and* if we can get it into Qt (properly with abi stable =
api) I would more than happily add support.<br></div><div><br>-----<br>Outp=
ut management protocol<br><br>Sway decided their users will just edit confi=
g files and reload compositor; but did say they&#39;d consider our protocol=
 if they add it.<br><br>------<br><br>Remote access protocol.<br><br>wlroot=
s has a draft  DMA buf sharing protocol, which is conceptually similar to o=
ur gem sharing but their protocol is much more...involved. It does a lot mo=
re but at the expense of being quite complex for both the compositor and an=
y potential recording client.<br><br>Does show plenty we could add to ours,=
 but we should see how stuff with pipewire pans out first and what&#39;s ac=
tually needed. <br><br></div><div>David<br><br></div></div></div>

--0000000000005bf4a9056b03d3d5--