Wayland Protocols Split Up From Weston, Wayland

[phoronix]

Phoronix: Wayland Protocols Split Up From Weston, Wayland

Jonas ?dahl announced the formation this morning of the Wayland-Protocols Git repository that will march to its own beat, separate of Wayland/Weston releases...

http://www.phoronix.com/scan.php?page=news_item&px=Wayland-Protocols-Repo

[bug77]

Not good. Not enforcing the availability of a reference implementation means we can (and probably will) have things in the protocol that look nice on paper, but are hard/expensive to implement.

[unixfan2001]

Sounds like we're well underway for X11 2.0 "The return of the unstable protocol".

[paintbynumbers]

There has been some concern recently on the Wayland development list that development has slowed and there are several items outstanding that need to be resolved for the desktop. This looks like it frees Gnome/KDE to develop new protocols, use them in their compositors, and try to standardize them in the common protocols repository (sure they'll remain unstable at first).

Perhaps he's sick of things like pointer confinement not going anywhere because nobody plays games in Weston?

[dh04000]

Sounds like we're well underway for X11 2.0 "The return of the unstable protocol".

X11-2: The Protocol Strikes Back

[bison]

Phoronix: Wayland Protocols Split Up From Weston, Wayland

Jonas ?dahl announced the formation this morning of the Wayland-Protocols Git repository that will march to its own beat, separate of Wayland/Weston releases...

http://www.phoronix.com/scan.php?pag...Protocols-Repo

What is the difference between Wayland and Wayland-Protocols? Weston is the reference implementation of one or the other, but what is the other?

[computerquip]

Wayland now is just the implementation of the protocol that lets you talk to the compositor. Weston is the compositor itself. Wayland-Protocols is the protocol being implemented as far as I understand.

[Trevelyan]

If I'd compare it to network protocols, I'd say:

Weston -> Reference implementation
Wayland -> IP/ICMP (and library implementation)
Wayland-Protocol
-> UDP/SNMP
-> TCP/HTTP
-> UDP/NTP

Or to the web world: Wayland is like JSON, and Wayland-Protocol details the sorts of messages you'd pass via Wayland/JSON. e.g. Messages about mouse gestures made: What sort of Key/Values would you expect, and how should you interpret those values.

[justmy2cents]

Not good. Not enforcing the availability of a reference implementation means we can (and probably will) have things in the protocol that look nice on paper, but are hard/expensive to implement.

except, if you read the reasoning for this, you'd not say anything like that. patches or changes were just bikesheded and not implemented, since it took way too many uninterested people to say "yes".

one of the statements in that previous talk about this was that only common interests can be implemented, aka. all interested desktops have to agree on it, but that is where it stops which means desktop developers decide what is needed. before that, people with access to approve changes sometimes didn't even understand the reasoning since their work is completely different topic had to and this led to standstill

also,. it is really hard to believe that desktop developers would create something that is hard to implement if it takes agreement from all interested parties

[sarmad]

Makes sense. We now have several implementations of Wayland (Gnome, KDE, Sailfish, Tizen), so there is no need to gate Wayland releases by Weston.