Announcement

Collapse
No announcement yet.

Mir Works On Screencasting, Parallelized Page Flipping

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

  • #61
    Originally posted by Onuca_V View Post
    Of course. Mir can't, anytime soon, be a replacement for X. It can be only a replacement for Ubuntu and other distros using Unity (are there any others?). If it's fine and not great, well that's a pointless discussion.

    I want Mir to be great, don't get me wrong. But I think Mir can't replace X in the GNU/Linux world. So, both have a place in this ecosystem: I'm just saying is clear Mir is for Unity/Ubuntu and an alternative to X needs to be developed for everyone else. If both projects succeed even better. I did not use the word "bad" at all.
    its my understanding that mir would would in any DE, ofc canonical wont make it easy but they wont make it hard either.
    they are solving theyr issues they arent gonna give you something that you can just copy&paste and be done with it, even wayland requires man power to work.
    all DEs have to do is want it enough just like wayland...

    Comment


    • #62
      Originally posted by TheSoulz View Post
      its my understanding that mir would would in any DE, ofc canonical wont make it easy but they wont make it hard either.
      they are solving theyr issues they arent gonna give you something that you can just copy&paste and be done with it, even wayland requires man power to work.
      all DEs have to do is want it enough just like wayland...
      And that brings up the big question. What does Mir do to make the other DE's want it? What does it offer that Wayland doesn't already offer?

      With Wayland, there's open development, where DE developers are welcomed to join in and contribute and write extensions and offer them to be merged to the core protocol and everyone benefits from the work of everyone. Everything is out in the open, there's no asymmetric CLA's (for those who care about such things, personally I've never seen it as a huge concern but it matters to some and thus it will alienate many developers like Upstart did) and there's no requirement to make Unity first, everything else secondary.

      Everyone can take Wayland and make it work for them, because if it doesn't have something they need, they can add it, but the same is not true of Mir - Mir only cares about Unity. If you wanted to use Mir and said "I need to extend the functionality because I need this thing for my DE" Canonical might say, "sorry, we don't need that for Unity" and that's that, no dice. With Wayland, you can always extend it yourself and if the extension is good, it'll get adopted in core protocol, but even if it doesn't, you can still use it in your own DE.

      Comment


      • #63
        Originally posted by TheOne View Post
        Lets see how running a gnome and kde application at the same time will be when wayland hits mainstream as how packagers will deal with the different wayland backends.
        This isn't a problems since both will be talking the Wayland protocol. It's like
        Http. No matter what web-server or web-browser you have, it works as both
        speaks a (de-facto) standardized protocol.

        Originally posted by dee. View Post
        So it's basically at the same point as Cyanogenmod at this point.
        Somewhere like that, yes.

        Originally posted by dee. View Post
        There aren't any compositors. Canonical's implementation is the only one.
        One year after Waylands announcement there only where Weston.
        The design pattern still applies.

        Originally posted by dee. View Post
        They "promise" lots of things... like they promised to support Wayland...
        They never promised that. As a developer I have yet to be let down. They
        deliver on schedule and keeps their promised support time. Also changing the
        API would upset a lot of people inside Canonical working on both Unity 8 and
        unity-system-compositor.

        Originally posted by dee. View Post
        Right, like libwayland.
        No not at all. libwayland just makes sure the server can talk to the clients and
        vice versa in a (de-facto) standardized way.

        Originally posted by dee. View Post
        Right, like what Wayland does.
        No.

        Originally posted by dee. View Post
        That's FUD and just plain wrong. There is a base implementation, libwayland.
        libwayland just maps the protocol to an API. Weston is a refferece implementation,
        however even the Wayland developers themselves says you shouldn't use it.

        Originally posted by dee. View Post
        Besides which, that whole "let's put everything in one monolithic beast" approach was already tried with X, and it wasn't good.
        Mir is nowhere near X. Mir is much closer to Wayland but with the difference that
        it is a Display Server and handles the hardware side for its compositors.

        Originally posted by Onuca_V View Post
        Second, Mir is not working just fine right now and I hope it will work fine by the end of the year (wishful thinking, maybe, let's see at the middle of the year)
        It works fine on the phone and tablet. No one have started to use it on the
        desktop yet.

        Originally posted by Onuca_V View Post
        Third, IF Mir works great this year, how much work would be needed for GNOME, KDE or E19 to implement Mir? Is it even viable?
        Less work than to implement Wayland.

        Originally posted by Onuca_V View Post
        What if Canonical changes Mir license just because they need to?
        They doesn't have a track record of doing that. And yet if they do, we'll see
        by then what license they change to. If it's bad we can always continue to
        use and fork those releases that are under GPLv3 and LGPLv3.

        Originally posted by dee. View Post
        What does it offer that Wayland doesn't already offer?
        Hardware abstraction

        Originally posted by dee. View Post
        With Wayland, there's open development, where DE developers are welcomed to join in and contribute and write extensions and offer them to be merged to the core protocol and everyone benefits from the work of everyone. Everything is out in the open, there's no asymmetric CLA's (for those who care about such things, personally I've never seen it as a huge concern but it matters to some and thus it will alienate many developers like Upstart did) and there's no requirement to make Unity first, everything else secondary.
        The CLA is a problem for developers yes, and we would probably never see mass
        adoption (partly because of Wayland and partly because of CLA). However for a
        user it doesn't matter at all.

        Comment


        • #64
          Originally posted by Pajn View Post
          They never promised that.
          No, they would never... wait, what's this: http://www.markshuttleworth.com/archives/551 ?

          Less work than to implement Wayland.
          Source? And if that's the case why do GNOME, KDE and E19 have Wayland backends but no Mir backends?

          Hardware abstraction
          Again: Source?

          Comment


          • #65
            Originally posted by Pajn View Post
            Somewhere like that, yes.
            While Wayland is being used on actual production phones. Phones you can buy with Wayland preinstalled.

            Wasn't one reason for Mir supposed to be that they need a faster solution for their mobile devices?

            One year after Waylands announcement there only where Weston.
            The design pattern still applies.
            So what you're saying is, it's going to be 5 years from now before Mir becomes viable enough for anyone to consider it? Ok...

            That's a really pointless comparison though, the two systems were made in very different circumstances. No one started developing a Wayland compositor one year after the work started on Wayland, because there would have been no point in doing so - most of the time spent in the beginning years of Wayland was doing the plumbing, the groundwork, preparing the graphics stack for a new, modern display system. To a layman observer it may appear like Wayland took many years to accomplish nothing, but in fact they did a lot in that time. A lot of things that were necessary to support a modern display system.

            Then when Wayland was done with the hard-but-invisible work, and got into the more user-visible stage - this is where the DE's and toolkits got into implementing support big time, causing more user-visible development, making it appear to the layman as if the development is "picking up speed" - at that point, Mir jumped in, took advantage of all that hard groundwork made by Wayland developers, and started building their own system.

            So Mir wouldn't even be possible without the groundwork done by Wayland developers, and its development only appears faster because they took advantage of all that work that was made for them.

            They never promised that.
            Yes they did.

            http://www.markshuttleworth.com/archives/551

            Originally posted by Shuttleworth in 2010
            We considered and spoke with several proprietary options, on the basis that they might be persuaded to open source their work for a new push, and we evaluated the cost of building a new display manager, informed by the lessons learned in Wayland. We came to the conclusion that any such effort would only create a hard split in the world which wasn’t worth the cost of having done it. There are issues with Wayland, but they seem to be solvable, we’d rather be part of solving them than chasing a better alternative. So Wayland it is.
            As a developer I have yet to be let down. They
            deliver on schedule
            Like the Ubuntu TV? Ubuntu on Android? Oh, and last summer they were saying how Ubuntu phones would be in stores by now. Don't see any Ubuntu phones in stores... And Mir itself is a pretty good example as well.

            Seems to me like Canonical makes lots of grand gestures and hypes up future products which they end up being unable to deliver. There's a word for that: vaporware. Now I don't want to be too harsh on Canonical, they are trying, but sometimes the company just seems like an attention-deficit child, starting new shiny projects everywhere, but never following through on any of them... they should pick something and concentrate on that, instead of trying to do everything at once and then later realizing they don't have the resources to actually follow through on their plans.

            and keeps their promised support time.
            That they have so far done, which is commendable.

            Also changing the
            API would upset a lot of people inside Canonical working on both Unity 8 and
            unity-system-compositor.
            So? Those people are being paid to work on whatever they work on. If they're upset, they're upset - doesn't matter, they work on what their boss tells them to work on. Many people get upset in and because of their work.

            Besides, Canonical has a history of upsetting some of their developers/contributors... remember the guy who was working on porting LightDM to Wayland? Who wasn't even told about Mir?

            No not at all. libwayland just makes sure the server can talk to the clients and
            vice versa in a (de-facto) standardized way.

            No.

            libwayland just maps the protocol to an API. Weston is a refferece implementation,
            however even the Wayland developers themselves says you shouldn't use it.
            Implementing a compositor in Wayland is very easy. The protocol API allows the developer to implement features that are already made possible by the API, and do all this in a compatible way.

            Besides, nothing stops anyone from forking Weston and using it as a base to build up their own compositor.

            Mir is nowhere near X. Mir is much closer to Wayland but with the difference that
            it is a Display Server and handles the hardware side for its compositors.
            Right, so everything is statically implemented, and if you want to support hardware Canonical is not interested in? Tough luck. And it's just going to end up in the same state X is now - with Wayland, the entire system is designed so that it can be extended and renewed but in a backwards-compatible way, preventing the same problem from happening as where we're now with X. With Mir, the one monolithic implementation is eventually going to get old and outdated, just like X did.

            It works fine on the phone and tablet. No one have started to use it on the
            desktop yet.
            And no one is selling phones or tablets with Mir preinstalled. Phones/tablets aren't like computers where you can have a viable business model by selling just the operating system. You NEED preinstalls for there to be any chance of making it in the mobile. Otherwise, you're just pouring money into a geek hobby.

            Less work than to implement Wayland.
            No. That is total and utter BS.

            They doesn't have a track record of doing that. And yet if they do, we'll see
            by then what license they change to. If it's bad we can always continue to
            use and fork those releases that are under GPLv3 and LGPLv3.
            That doesn't matter and isn't the point. I don't care about the CLA. The point is that many developers don't see it that way, they see the CLA as asymmetrical and suspicious, and are discouraged from wanting to contribute.

            Hardware abstraction
            Wayland offers that too.

            The CLA is a problem for developers yes, and we would probably never see mass
            adoption (partly because of Wayland and partly because of CLA). However for a
            user it doesn't matter at all.
            It matters to the user when the software they want to use isn't compatible.

            Comment


            • #66
              Originally posted by dee. View Post
              While Wayland is being used on actual production phones. Phones you can buy with Wayland preinstalled.
              Sure. Never said against that, never will. Doesn't really care however. I don't like the Tizen UI
              (this is very personal, I'm not saying it's bad). However this doesn't make Mir a lesser alternative.

              Originally posted by dee. View Post
              Wasn't one reason for Mir supposed to be that they need a faster solution for their mobile devices?
              Not really, it were that mobile needs server allocated buffers. And no one does that in Wayland
              (except for Tizen) and as they want mobile and desktop to use exactly the same codebase
              they would need to use server allocated buffers on desktop as well, creating troubles for
              stuff expecting client allocated buffers. Having a hard breakage is usually better as you then
              can handle that in a good way (add Mir support in toolkits or a WaylandMir layer), instead
              of accepting random buggs and breakages.

              Originally posted by dee. View Post
              So what you're saying is, it's going to be 5 years from now before Mir becomes viable enough for anyone to consider it? Ok...
              Probably not 5 years because, as you said, Wayland started in a totally different position.
              However more than a year (and especially more than ~6 months of actually working) should
              be expected.

              Originally posted by dee. View Post
              Then when Wayland was done with the hard-but-invisible work, and got into the more user-visible stage - this is where the DE's and toolkits got into implementing support big time, causing more user-visible development, making it appear to the layman as if the development is "picking up speed" - at that point, Mir jumped in, took advantage of all that hard groundwork made by Wayland developers, and started building their own system.
              I'm not denying that work and I'm really thankful. Else we would be stuck with X for several years.

              Originally posted by dee. View Post
              Yes they did.
              I'm hardly calling that a promise, just publish a decision. Decisions are subject to change.

              Originally posted by dee. View Post
              Like the Ubuntu TV? Ubuntu on Android? Oh, and last summer they were saying how Ubuntu phones would be in stores by now. Don't see any Ubuntu phones in stores... And Mir itself is a pretty good example as well.
              No API or developer information were published for Ubuntu TV or Ubuntu on Android so neither
              I nor anyone else have work that is on hold because of that. At the time an SDK were released
              for Ubuntu Touch is were known not to be available at spring. If no phones are released this
              year we can start talking about a delay.
              For stuff that matters to developers, like distro release. The last delay were 8 years ago.

              Originally posted by dee. View Post
              So? Those people are being paid to work on whatever they work on. If they're upset, they're upset - doesn't matter, they work on what their boss tells them to work on. Many people get upset in and because of their work.
              True, however that would just be extra work for themselves. I don't have any reason to break
              the API on purpose. As they don't promise the protocol they can easily make changes that
              doesn't break API and just add new headers if necessary.

              Originally posted by dee. View Post
              Besides, Canonical has a history of upsetting some of their developers/contributors... remember the guy who was working on porting LightDM to Wayland? Who wasn't even told about Mir?
              Yes, that were very badly handled.

              Originally posted by dee. View Post
              Implementing a compositor in Wayland is very easy. The protocol API allows the developer to implement features that are already made possible by the API, and do all this in a compatible way.
              It's not easy if you want one in pair with Unity, Gnome, KDE, Xfce or similar. Of course that isn't
              easy on Mir either, but should certainly be easier.

              Originally posted by dee. View Post
              Besides, nothing stops anyone from forking Weston and using it as a base to build up their own compositor.
              Not stopping, but the Wayland developers themselves recommend against that. Weston is only
              created to test the Wayland protocol as it develops, it's not intended for production.

              Originally posted by dee. View Post
              Right, so everything is statically implemented, and if you want to support hardware Canonical is not interested in? Tough luck. And it's just going to end up in the same state X is now - with Wayland, the entire system is designed so that it can be extended and renewed but in a backwards-compatible way, preventing the same problem from happening as where we're now with X. With Mir, the one monolithic implementation is eventually going to get old and outdated, just like X did.
              First, X is very extendible. That is why it have been able to survive for this long. Without stuff
              like DRI and DRI2 we wouldn't have managed this far.
              Second, Mir also support extensions. But it does the job that every compositor would want to
              do anyway.

              Originally posted by dee. View Post
              And no one is selling phones or tablets with Mir preinstalled. Phones/tablets aren't like computers where you can have a viable business model by selling just the operating system. You NEED preinstalls for there to be any chance of making it in the mobile. Otherwise, you're just pouring money into a geek hobby.
              Of course. That is why Canonical have signed deals with a number of big operators.
              http://www.ubuntu.com/phone/carrier-advisory-group
              I can only talk for 3 as those are the only one with business in my country but they
              are a carrier that often launch new thing. Here they were the one launching 3G,
              Feature phones (camera, video talk, mp3 and such) and Android. I would actually
              be more surprised if they didn't try to launch Ubuntu than if they do.

              Comment


              • #67
                Originally posted by Pajn View Post
                Not really, it were that mobile needs server allocated buffers. And no one does that in Wayland
                (except for Tizen) and as they want mobile and desktop to use exactly the same codebase
                they would need to use server allocated buffers on desktop as well, creating troubles for
                stuff expecting client allocated buffers. Having a hard breakage is usually better as you then
                can handle that in a good way (add Mir support in toolkits or a WaylandMir layer), instead
                of accepting random buggs and breakages.
                No. The decision between server side and client side allocated buffers is done entirely in the EGL implementation, Wayland and Wayland compositors don't care, at all. You can use the same compositor and toolkits with the Mesa libEGL, which uses client side buffers or with some other libEGL which uses server side buffers, nothing will break.

                Comment

                Working...
                X