Reassessing Wayland

So somewhat less than 3 years ago, I made my Infamous Wayrland Rant post most likely the most readable post on this blog. I should really write about music again one of these days, but that is a subject for another time. Speech maybe a little inflammation, but I feel the criticisms I make in the time fair. Mainly born some failures I had in full ecosystem, and it wasn’t like I was the only voice. There are other people there you can find that they meet their own land problems and write them about it.

With that post, I probably rejected myself as an anti-Wayland man I did, but I promised you not the situation. You can check my MPV practices, and these businesses are like usual. Many road repairs, parts, and all good things. Many hours ago since then, and it’s over to see the situation with all new developments in mind. To be Frank, my original post is better and unfairly to leave it in the current state that does not recognize the work done. So compared to 3 years ago, I have more positive sight now.

The entrance to the upper ground fixed to most of the technical things I complain.

I mean honestly this kind of value. In retrospect, it’s a bit odd. I may need to complain to the public when I get these kinds of results (it is a joke of all). But a good example here is the support of explicit sync. It’s not too long, I don’t think we’re going anywhere close to it something. Frank, Linux is generally farther away from this front. Getting clear syncs get a lot of work across graphics space: kernel, table, protocol changes to all companies (a famous playue of all companies Put on work to get it. I honestly consider we will still sync with most of the 2020s.

Another great development is to introduce committee and FIFO protocols at the end of resolving the feared “composer of the rending thread” issue. Previously, Wayland clients had no way to see if they appear and composers do not always suffer from your client’s interpretation if you haven’t been offscreened. For most, the workaround in this case is to set up mailbox mode (or equal to whatever you do) and just burst. People who raise issues with this behavior of draining often with equal answers, so I doubt it will change.

Today now supports Command-timing and each of the vulkan allowing surfaces to remain at Vsync with no compositor composer. It’s more important for people playing wine games and what. It is not like refacting the translation of ancient proprietary games that can be a way friendly to the land that can, so this behavior is a brick wall. Composer’s implementations are still in the works for it and no EGL side support, but from my attempt to work well. From the view of the MPV, we are allowed to skip a specified piece of code I personally found cursed, so I am happier. It is also possible to use the time and FIFO in the room for some of our other Wayland Vos. That requires investigation.

Referring to the table, I broke it slightly to the old post, so I think I need to apologize today. It’s true that in time, implementations are little in the lack of side, but this story is different now. In 3 years since then, a large amount of effort was placed in Wyland WSI. Many developers make some changes change. Items are like easy mode, clear sync, timestamps to present timestamps, and others are there today. There is also an open mr right now for the color management protocol for the vulkan on the table. That protocol is in development Hell for a long time, but we all finally see the light at the end of the tunnel. I expect it to end in a future. There are several proven implementations, and only the only comments are left all minor. Once in place, HDR on Wayland just works in the box. It’s fun honestly.

Wayland-protocols in upstream itself have better.

This is not exactly a secret, but ground protocols are a painful stream. No one of the main people is not pleasing or anything; I have a positive experience with myself. This is more that that movement is generally in that space slowly slow and every little thing is mentioned in death. The bar for picking up protocols higher, and it seems to be something almost impossible.

I’m not sure if there is a definite turning point, but things in that repo will work harder than before. There is a relatively high united request to add an experimental namespace to land protocols. Of course, it is welcome, but honestly even before I think repo has made jumps and boundaries to review and accept contributions. If I made my original post, Wayland protocols were at 1.25. At the time of writing, we’re in 1.40, so a whopping 15 more versions. And it’s not like those are just too little of no important bumps. Many protocols I would consider most (Fractional-Scale, Linux-DRM-SNTCOBJ, with long-time control. One of the regeland complaints is what you have with your feelings with your feelings Sometimes and protocols like that far in response to that concern.

Heck even small things are better now. I never complained about it because it wasn’t a big deal, but I always thought that the default stance on the roads of the client’s sides. Most applications ignore cursor drawing. Considering most composers draw a cursor, why aren’t these clients used? And actually it is set to the cursor form protocol. All the little annoyance about the unequal cursor sizes to go whether you only allow compositor. Again, not a primary issue or anything but it is a good example of one of the little things mentioned. In addition, there is either a screencopy protocol again. This was another one of those things I thought that was going to be a “never ever” since many compositor developers were not particularly found of the idea of ​​client potentially recording the screen. Now sure, many compositors do not support it today, but in the end you can target something through Wayland APIs to get your need.

Of course, nothing is perfect.

Then, I feel a little annoyed with people overselling Wayland when there are many things wrong with this at the basic level. But this story is different today. For most use of cases, Wayland should be as good or better at this point. There is a big list of reasons you need to weigh to make this analysis, but scales are in favor of the ground. Now of course, there is a lot of work to do. A personal pain in myself the inability is the inability to embees foreign surfaces without making your application of self composer. It is one of the things that work well on other platforms, but still lacking the road on the ground. Another is the full issue of an application with many windows in charge of them on the ground. That’s a pretty visible issue when it is enlarged, and the job still makes trying to find a solution for that.

In terms of development philosophy, the custom group of compositors / upstreams is more than the “policy” part of the fence opposed to “mechanism”. I personally dependent on the opposite, but the road on the way home gets easier easily. There are so many things you used to be unable to do it now. As a small point of pain, the overall area of ​​the Ecosystem land is still a mess. You have soldiers with composers to deal with and the phthora of thumb protocols. This is just a result of the core protocol that is very small and everyone wrote their own server implements instead of sharing one, but manageable. I’m sure to write a boilerplate set for any protocol I want to use and try against the usual group of composers if this is my need, but what I need, but if it’s a happy camping to the end of the day.

So what about Xorg?

The story is a little interesting here since 2024 saw an influx of xoror activity due to a person even if it may end a fluke. Time will tell one. There are people who quickly screamed Xorg, but I still think it has its place and there are many users of this (not just Xwayland) for many years to come. Sure, GTK 5 doesn’t support Xorg not surprisingly, but I doubt it or most of the applications overall watching their Xorg backends at any time. And there are some more informed gaps between Xorg and Wayland compositors. For example, the newly informed of the differences in the mouse design is true and something I noticed but the difference is not more blame. Something like this is just one of the inevitable results of Wayland’s design of different kinds.

I still want the place to be a little conservative and “xorg-like” in its original approach. Of course, it’s too late for it, but it feels so many causes of pain points is self-inflicted and can avoid less self design choices. But at the same time, it’s true that the place where all work goes on these days and tell me that it makes xorg with most things at this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point in this point. Like for Xorg, if people still want to grow and use it, I don’t see that problem. It can only be together. Perhaps a day in the future I know that almost no one in Xoror Code. If we reach that point, I can consider removing the code. But as far as the future, and I have no reason to worry about anytime.

Meanwhile, I will be fair to developers in the land and composer. They are healed to your of things just new and should be honored for it. On the side of the MPV at home, I helped several times and got good contributions from many developers in the composer. It’s a long carrier, but real things in better situations today. The amount of progress in a short period of time is unique and truly unexpected. Invalid to continue to grow and negatively when real problems can be resolved. Let’s see how the rest of the decades, but it’s like an upward slab for so far.