Blumenkrantz Seeks Clear Policy How Wayland Protocol Changes Can Be Rejected

[phoronix]

Phoronix: Blumenkrantz Seeks Clear Policy How Wayland Protocol Changes Can Be Rejected

As part of his new hope for helping to accelerate Wayland protocol development, Mike Blumenkrantz with Valve proposed an "experimental" protocol development area within Wayland-Protocols. He's also laid out a proposal for seeking to solidify the means by which suggested protocol changes can be rejected...

https://www.phoronix.com/news/Blumekrantz-Wayland-NACK

[kiffmet]

Good. It would be very appreciated if the whole ordeal of getting potential new Wayland protocols evaluated and pursued/dropped was streamlined and accelerated.

[access]

Looks quite sane if you actually read the proposal. Could perhaps be adopted by the Linux kernel too vis-a-vis the latest Rust brouhaha.

[clippy]

Skimming through several stalled MR discussions on wayland-protocols, I think there is a running theme of conflicting interests between wayland implementers on the server side (compositors) vs on the client side (toolkits, mesa, wine). Currently, the list of wayland-protocols members consists mostly of the former. While clients would like to see more features implemented faster, the server developers advise caution, as not to push out half-baked solutions and/or repeat past mistakes with X11. As I see it, this makes it very easy for proposals to stall, as the skewed proportion of server-side devs increases the likelihood of someone finding something to complain about in any given proposal.
Also, server-side projects themselves have no inherent stake in pushing forward with protocols that only benefit clients, as to them, such protocols only mean that there's more complexity to implement. This leads to many such proposals running out of steam due to developer inactivity.
As I see it, these proposed restrictions on NACKs would address these problems very elegantly. Combined with the admission of Mesa into the list of wayland-protocols members, my fingers are crossed that the current deadlock is soon going to be history.

[pdbecid]

Mike is doing God's work. Never thought someone would tackle the wasp nest that is wayland-protocols development so elegantly.

[Britoid]

Skimming through several stalled MR discussions on wayland-protocols, I think there is a running theme of conflicting interests between wayland implementers on the server side (compositors) vs on the client side (toolkits, mesa, wine). Currently, the list of wayland-protocols members consists mostly of the former. While clients would like to see more features implemented faster, the server developers advise caution, as not to push out half-baked solutions and/or repeat past mistakes with X11. As I see it, this makes it very easy for proposals to stall, as the skewed proportion of server-side devs increases the likelihood of someone finding something to complain about in any given proposal.
Also, server-side projects themselves have no inherent stake in pushing forward with protocols that only benefit clients, as to them, such protocols only mean that there's more complexity to implement. This leads to many such proposals running out of steam due to developer inactivity.
As I see it, these proposed restrictions on NACKs would address these problems very elegantly. Combined with the admission of Mesa into the list of wayland-protocols members, my fingers are crossed that the current deadlock is soon going to be history.

I predict no one on that Wayland member list plays games on Linux, because then they'd realise the status quo is not workable and that their decisions are having a negative effect on the adoption of Wayland.

Consumers like Mesa, SDL, Valve need to be brought on as these are the people fucked over when dumb decisions are made.

[jeisom]

I am glad there is a push to make changes and improvements happen faster. If they don't already, it would probably benefit picking a compositor as a testbed for new protocols sort of how nvidia has a vulkan beta driver for testing and the release drivers for vulkan support. Protocols could then be implemented and tested there before migrating to the others. I am not super familiar with the process the w-p devs use other than it is moving very slowly though.

[Weasel]

Mike is doing God's work. Never thought someone would tackle the wasp nest that is wayland-protocols development so elegantly.

His MR that changes the whole thing might get NACKed lol.

[fitzie]

they are treating wayland-protocols like the un security council, everyone get's a veto and rarely does anything get any consensus. it should be more like an incubator. anyone with an idea is encouraged to discuss it, implement it, receive feedback. and if it has merit then it becomes promoted to one of those holy namespaces they once enough compositors implement it. nothing should automatically get into the protected namespaces just because the cabal wants to push it through, nor should something that's implemented by some large subset of compositors be held back because the cabal wants to bikeshed it for a decade or two.

[Errinwright]

I have to admit that I am surprised this type of proposal wasn't made earlier by [anyone]. Satisfied with a Valve employee taking up arms for the gaming community.