Announcement

Collapse
No announcement yet.

The Technical Plans For Making Wayland 1.0

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by TheBlackCat View Post
    Uh, why exactly?
    In my opinion, it doesn't really solve any problems that need solving, while introducing negatives.

    It is a pointless break with the past. I have no real issue with breaking from the past if there is a good reason. For example I support PA, and Systemd looks like it will be good (I need to find more rebuttals from upstart devs, though. But either way it is leaving sysvinit behind). But when it gains you nothing...

    And you lose things as well.

    Network transparency - sure, they say "we'll just layer it on top, it will be easy and work well". But so far it is a no-show. I also haven't seen whether or not it would be possible to do single windows like X, as opposed to full desktop.

    Then there is the client-side window decoration thing, which is just stupid.

    And the "you have to write a compositor to write a WM" thing seem iffy to me. I am not sure how big a problem this is, but it was not reassuring the way they first talked as if there was no possibility of WMs. Just a misunderstanding, it seems, but the compositor may still be an issue.

    And as I said, it breaks with the past - how many wonderful X WMs will not be ported/recreated for Wayland? I don't like it, no sir.

    A few times I have wondered if I just am not hearing the wonderful things it will bring instead, but then I have gone looking, and I have not found any truly inspiring essays/discussions about Wayland.

    Comment


    • #12
      Originally posted by KellyClowers View Post
      In my opinion, it doesn't really solve any problems that need solving, while introducing negatives.

      It is a pointless break with the past. I have no real issue with breaking from the past if there is a good reason. For example I support PA, and Systemd looks like it will be good (I need to find more rebuttals from upstart devs, though. But either way it is leaving sysvinit behind). But when it gains you nothing...

      And you lose things as well.

      Network transparency - sure, they say "we'll just layer it on top, it will be easy and work well". But so far it is a no-show. I also haven't seen whether or not it would be possible to do single windows like X, as opposed to full desktop.

      Then there is the client-side window decoration thing, which is just stupid.

      And the "you have to write a compositor to write a WM" thing seem iffy to me. I am not sure how big a problem this is, but it was not reassuring the way they first talked as if there was no possibility of WMs. Just a misunderstanding, it seems, but the compositor may still be an issue.

      And as I said, it breaks with the past - how many wonderful X WMs will not be ported/recreated for Wayland? I don't like it, no sir.

      A few times I have wondered if I just am not hearing the wonderful things it will bring instead, but then I have gone looking, and I have not found any truly inspiring essays/discussions about Wayland.
      Well, most of your points say "I just want it to die because I will not use it". If you are OK with using an unmaintained WM (if it's being maintained, I don't know why do you suppose it will not be ported over to Wayland if it becomes mainstream), why wouldn't you be OK with using an unmaintained (which in fact, will not be unmaintained, just will probably not be used by default to end-user distros) display architecture.
      You may not see the benefits, but I really doubt the people who developed it wants to lose time, so I assume there are. They're just useless for you, maybe you are not on the ideal target of it. That still doesn't explain why would you want it to die. Don't like it, don't use it.
      Personally, I'm only interested in fast start and fast rendering. About fancyness, the most heavy effects I use are shadow and transparency, and I can live without it, and both Wayland and X.org support all I use. I'll use the faster that supports any lightweight DE. If it doesn't suit your needs (like being able to use an older state of the art WM), you still can use X, you know. No need to wish a project which suits someone else's needs to die.

      Comment


      • #13
        >Well, most of your points say "I just want it to die because I will not use it". If you are OK with using an unmaintained WM (if it's being maintained, I don't know why do you suppose it will not be ported over to Wayland if it becomes mainstream), why wouldn't you be OK with using an unmaintained (which in fact, will not be unmaintained, just will probably not be used by default to end-user distros) display architecture.

        If Wayland achieves dominance, X will be minimally maintained or unmaintained (or rather, one, then the other). No one is going to put that much effort into it. In that case X cannot survive long term any more than the KDE 3 and Gnome 2 forks can. Sure, I might be able to use it for a while, but eventually too much stuff will be broken.


        >Personally, I'm only interested in fast start and fast rendering.
        Those are ok goals, I don't think Wayland is worth it just to get there.

        Comment


        • #14
          Originally posted by KellyClowers View Post
          >Well, most of your points say "I just want it to die because I will not use it". If you are OK with using an unmaintained WM (if it's being maintained, I don't know why do you suppose it will not be ported over to Wayland if it becomes mainstream), why wouldn't you be OK with using an unmaintained (which in fact, will not be unmaintained, just will probably not be used by default to end-user distros) display architecture.

          If Wayland achieves dominance, X will be minimally maintained or unmaintained (or rather, one, then the other). No one is going to put that much effort into it. In that case X cannot survive long term any more than the KDE 3 and Gnome 2 forks can. Sure, I might be able to use it for a while, but eventually too much stuff will be broken.


          >Personally, I'm only interested in fast start and fast rendering.
          Those are ok goals, I don't think Wayland is worth it just to get there.
          With the first point, I'm pretty sure at least all the other OS's devs would try to maintain X, since Wayland depends heavily on an infrastructure only Linux currently provides. Also, I don't think it will be dropped while it doesn't provide equal or better features. As far as I know (not from first hand, sadly), network transparency is a real need for sysadmins, so until Wayland achieves it, I don't think X can be dropped.
          About broken-ness (I don't think that's a real word, but I'm too tired to think of a proper term), it should keep running, the only problem you will probably have is the lack of new software (of course, at the point (and IF that happen) Wayland is predominant), but being completely on user land I think should suffice while there is no major general API breaks. You will lose direct rendering in new kernels if/when a big change on the graphics stack happens kernel side.

          About my goals, I agree, I wouldn't develop a brand new server just for that goals, but being happening, I'd chose which I'll use in that basis.

          EDIT: Also, with older unmaintained WM's, you'll get to the point where everything break itself, so your points contradicts each other. Another example is if there is an X12 protocol. There will be a break with the past at some point. I agree that maybe Wayland's case has no enough reasons for that breakage, but it will happen at some point.
          I don't like to see how things get dropped either (in most cases, though, it's because there are no maintainers, so I can't blame anyone else than me by not being a dev ).
          Last edited by mrugiero; 25 February 2012, 12:15 AM.

          Comment


          • #15
            Originally posted by KellyClowers View Post
            In my opinion, it doesn't really solve any problems that need solving, while introducing negatives.
            How much low-level X11-based development do you do? I don't do any, so I rely on the people who do to tell me whether wayland will help their efforts or hurt them. They seem pretty overwhelmingly in support of wayland from what I have seen, and I tend to trust their judgement.

            Comment


            • #16
              Originally posted by KellyClowers View Post
              In my opinion, it doesn't really solve any problems that need solving
              It solves several problems that need solving. That's why it exists in the first place.

              Network transparency - sure, they say "we'll just layer it on top, it will be easy and work well". But so far it is a no-show. I also haven't seen whether or not it would be possible to do single windows like X, as opposed to full desktop.
              It is trivially possible to do per-window remoting like X11 does with the Wayland design. It is also trivially possible to do whole-desktop remoting, if that's what you need. The protocol is agnostic to remoting and lets the remoting proxy do whatever it wants however it needs in whatever way the users want it to do.

              I've written about it before, sick of repeating the full spiel about why whining about Wayland's remoting is silly. It's possible. Nobody's done it because nobody who actually matters has cared enough to implement it. If you care about it, implement it.

              Then there is the client-side window decoration thing, which is just stupid.
              It is superior to server-side window decorations. It fixes several otherwise impossible-to-fix problems. At worst it introduces the "locked up windows can't be moved or killed" problem, but then only if the compositor is retarded and doesn't deal with dead zones and non-responding windows in a sane way. It's a non-problem in the real world, and all the people bitching about it are only bitching because they used Windows 95 once some years ago and saw problems that no longer affect any of the more recent modern composited client-side-only OSes, which Wayland is merely trying to catch up to in terms of user experience and feature set.

              And the "you have to write a compositor to write a WM" thing seem iffy to me.
              Not having a compositor is a problem. A serious problem. Not only does the lack of compositing bar a large number of useful features for users, it also massively complicates the entire graphics stack, the driver stack, rendering APIs like OpenGL (which still assumes a lack of a compositor, and hence is a total bitch to use when setting up contexts, because you have to directly bind to a shared framebuffer with a fixed set of color, depth, and stencil buffers, rather than each app getting its own configurable buffers; and so you end up being wasteful and using an extra FBO that you shouldn't need to have just to support crappy desktop platforms like Win9x/XP and Linux/X11).

              I am not sure how big a problem this is, but it was not reassuring the way they first talked as if there was no possibility of WMs. Just a misunderstanding, it seems, but the compositor may still be an issue.
              There is no possibility of WMs as you know them. They would be plugins to the compositor, or replacement compositors, at most.

              Nothing wrong with that, either. The world does not need 400 different broken WMs when just one good WM does all you need.

              Comment


              • #17
                It is superior to server-side window decorations. It fixes several otherwise impossible-to-fix problems. At worst it introduces the "locked up windows can't be moved or killed" problem, but then only if the compositor is retarded and doesn't deal with dead zones and non-responding windows in a sane way. It's a non-problem in the real world, and all the people bitching about it are only bitching because they used Windows 95 once some years ago and saw problems that no longer affect any of the more recent modern composited client-side-only OSes, which Wayland is merely trying to catch up to in terms of user experience and feature set.
                It happened to me just last week on Win7. Do tell, which "more recent modern composited client-side-only OSes" are you referring to?

                Comment


                • #18
                  Originally posted by elanthis View Post
                  It's a non-problem in the real world, and all the people bitching about it are only bitching because they used Windows 95 once some years ago and saw problems that no longer affect any of the more recent modern composited client-side-only OSes, which Wayland is merely trying to catch up to in terms of user experience and feature set.
                  Yes, I am sure developers of popular window managers know absolutely nothing about designing window managers

                  Comment

                  Working...
                  X