I never understood why Valve insisted on writing steam directly against the X api, rather than using a toolkit. It's going to cause them pain whatever happens with wayland and mir now.
If that is indeed the case, then all this ranting I've been doing about fragmentation was pointless and wrong. Please somebody give me proof that my ranting was indeed pointless and wrong. Not many times have I wished that I was wrong, but this is certainly one of them.
I think that nowdays many companies plan to stay with X no matter what will happen with Wayland and Mir, but when Wayland will start to show the big advantages against X, many people will change their mind.
Wayland is asyncronous by design, no blocking (only the client that suffers a problem stop to rendering), and all the others advantages that all know by the previous articles.
Yestarday I was reading on Mertin's blog (Kwin mainteiner) that with wayland he can debug Kwin inside the development session, instead collect debug's info by a different vt process. He said that this issue boring him by years.
That surprise me a lot. Probably, during the wayland adoption process, we will discover many more advantages that are not obvious just reading the architecture.
About Valve, why stay on X if you have a better choice?
When will be avaiable the binary blobs for Wayland, if you care about performance, if you care about a "every-frame-is-perfect" (each frame have a timestamp), if you care about a modern input system, if you care about an asyncronous display server, if you care about avoiding stupid bottleneck, why still stay on X if you have a so valid alternative?
Probably, even a legacy program can run better on xwayland than X, at the end of the day. :-)
Valve doesnt have any choice but to stick with X. If they move to Wayland they screw Canonical. If they move to Mir they screw everyone else. What's unfortunate is that if Mir never existed this wouldnt be an issue at all. There is already too much fragmentation. What Mir adds pushes it past a tipping point. Canonical should be forced to die. (yes, a.k.a murdered.)
SDL3 have to support Wayland AND Mir anyway. Since Valve applications is SDL-based what the problem for Valve to support Wayland and Mir?
Well, of course. Valve still needs to keep a solid marketing front. I'm pretty sure at least that much is obvious. The dumbass shit that Canonical is doing, I'm also sure is pretty obvious. Sooner or later Canonicals partners are going to get worn thin.