Slashdot Mirror

← Back to Stories (view on slashdot.org)

Mir Won't Ship Even In Ubuntu 14.04

Posted by Unknown on from the long-live-x11 dept.

jones_supa writes "As can be recalled, Mir didn't make it to the Ubuntu 13.10 release to replace X.org as the display server. Back then it suffered of problems in multi-monitor support, along with other issues. Now it turns out that Canonical's product will not make it even into the next LTS version (14.04) of the Ubuntu desktop. Mir itself would be ready for showtime in the schedule, but there are problems with XMir, which is the X11 compatibility layer that ensures Mir can work with applications built for X. The comments came at the Ubuntu Developer Summit: in an online event Mark Shuttleworth stressed that the 14.04 desktop has to be rock-solid for customers with large-scale deployments, such as educational institutions. In the meantime, you can already try out Mir in your Ubuntu system."

16 of 111 comments (clear)

  1. Interesting by ifiwereasculptor · · Score: 5, Funny

    I think Mir might eventually replace X. It's already been replacing Hurd for quite some time.

    1. Re:Interesting by ledow · · Score: 5, Insightful

      I honestly don't care what they use.

      The fact is, if I can program against the X libraries, and load up that have been programmed against the X libraries, and distribute programs that have been programmed against the X libraries, it needs to all "just work". And it needs to work as good as, or better than, X itself.

      When you have that, it pretty much doesn't matter what faddy crap is underneath.

      It's like if they replace ext2 with ext3, or ext3 with ext4. I don't care so long as I have tools to read the data, it works as a filesystem, and it has no downsides compared to the previous version. What my filesystem actually *is* is irrelevant so long as it works. What my display manager actually is is irrelevant so long as it works. And in the case of an X-compatible display manager, it has to work like X in all cases without me needing to make changes to my software.

    2. Re:Interesting by Mitchell314 · · Score: 2

      I'm just waiting for when Canonical announces their own kernel.

      --
      I read TFA and all I got was this lousy cookie
    3. Re:Interesting by DrXym · · Score: 3, Interesting

      So it's really baffling to me why Cannonical is reinventing the wheel here

      I think the main reason is by owning the display tech they can shut their competition out of the mobile / tablet space. Mir is dual licenced - proprietary or GPLv3. They can do what they like under their proprietary licence while hamstringing their competition with the onerous requirements of the GPLv3 (e.g. not being able to link proprietary drivers, limits on DRM). It's probably why Intel walked away from the project.

      The Ubuntu wiki is tries to provide technical reasons but it's all pretty vague - it alludes to issues like lack of support for 3d input devices and a few other problems with the protocol but these don't sound very convincing arguments to write everything from scratch.

    4. Re:Interesting by Microlith · · Score: 2

      I didn't say that Qt wouldn't have a Mir backend, I said they likely wouldn't accept it upstream. Fucking learn to read, drinkypoo. They're developing all of the Mir backends and will likely be stuck doing out-of-tree maintenance or all of it.

    5. Re:Interesting by unrtst · · Score: 2

      While I agree in spirit to much of what you're saying, I don't believe the requirements are as tight as you state, and the filesystem analogy is weak.

      While most linux distros (maybe all) default to using an ext variant, they also work with many other filesystems. Tools have been updated to support multiple filesystems, and others remain filesystem-specific. The same could happen for Mir/X/Wayland.

      Consider Mac OSX for a minute. It can run any X application after some re-compiling, but its native window manager doesn't do it. It's X support is not as good as a native X server, but it's good enough. Most popular apps have been patched or forked to support the native gui stuff. If QT and GTK gained Mir and Wayland compatibility, most apps could be recompiled to render directly to Mir/Wayland. For the edge cases, using a X compatibility layer that isn't quite as good would probably be just fine... at the very least, most users won't care or even notice.

      The network transparency part has been discussed in depth. In the end, there needs to be a way to export a display from a remote machine to the local machine, which the compatibility layer will likely handle without a problem (any advanced features like 3d are unlikely to work well in those situations anyway). And there needs to be a way to export a local app to a remote machine - this will probably require using a different tool/protocol on the receiving end (ex vnc), but that's going to be sufficient for most of those edge cases that needed that as well.

      Mir and Wayland will need a way to support X applications, but if they get proper support from the toolsets, then it doesn't need to be better, or even on par. As long as it works well enough and is stable, that'd be enough to start migration. I don't think either is there yet though.

      Personally, I'd like to see someone implement the opposite - get Wayland running on top of X. It'd be pointless in the long run, but it would let the user choose which gets first class support on their system (ie. they could still run a mix of apps using both, and app developers don't have to worry about how few people are on Mir/Wayland). I'd be surprised if someone out there hasn't toyed with this yet, and maybe even demo'd it.

  2. Makes sense by Anonymous Coward · · Score: 2, Interesting

    Taken on its own, it does make sense. LTS needs to be usable (technically, inb4 "unity") on the widest practical range of hardware and be supported for 3 years. If Mir needs to be delayed so X applications can run on 14.04, so be it.

    1. Re:Makes sense by rotaryexpress · · Score: 4, Insightful

      Supported for 5 years, not 3. But your point is valid. Mir can be rolled out in 14.10 (the next non-LTS release) and have a year and half of testing before the next LTS.

    2. Re:Makes sense by kick6 · · Score: 2

      Taken on its own, it does make sense. LTS needs to be usable (technically, inb4 "unity") on the widest practical range of hardware and be supported for 3 years. If Mir needs to be delayed so X applications can run on 14.04, so be it.

      I see it this way too. It's not, necessarily, an indictment of Mir. It might just be that, unlike some major OS vendors, canonical is taking a more measured approach than to push features that don't work into software they're going to have to support long term.

    3. Re:Makes sense by occasional_dabbler · · Score: 2

      Ubuntu LTS releases are now supported for five years even on the desktop since 12.04

      --
      "Our opponent is an alien starship packed with atomic bombs," I said. "we have a protractor"
  3. No kidding? by m2 · · Score: 4, Insightful

    You have the hubris to say that you are going to fix everything that is wrong with X11 / X.org AND also provide a compatibility layer on top of your new shiny solution to support running the programs that still use the thing you are claiming to fix ... and now you are surprised because getting said compatibility layer right turned out to be thornier than you had expected?

    Several years ago I wrote a transport mechanism on top of VNC that allowed you to access high end graphics services (read OpenGL) from devices without any hardware acceleration to speak of (back then it was an ipaq). I did the initial implementation in one evening, which worked for 80% of the use cases. Together with another developer, it took us probably a month to get it to 90%. A third party worked for half a year to get it to 95%. Several years later it was up to 98%... maybe.

    Whenever you try to pull this kind of stunt off, you are going to run into the same situation. Most of the stuff that you are interested in is easy. Then there's the stuff that makes "creative" uses of existing APIs. And then there's the stuff that works because of, not despite of, existing implementation bugs. And then you run into the really weird...

    1. Re:No kidding? by drinkypoo · · Score: 2

      No, no you didn't address why vnc doesn't cover those use cases. You made some excuses, but it works fine in most of them, or is almost adequate. I've been running X remotely since before vnc existed. You have no idea whatsoever what I've done, and I have plenty of perspective on the way X remoting is used. Further, I read your shitty comment, and I thought it was stupid but I chose to ignore the parts I thought were stupid and simply write a simpler comment that explained precisely what the problem was. But since you're being such a whiny little baby about this, let's go ahead and go back and see what you said, and why it is stupid. And in fact, it will also reveal why this comment of yours is stupid, and hopefully establish a pattern of stupidity so that when I call you an idiot I am modded insightful instead of troll or flamebait.

      Let's begin.

      VNC being 1 to 1 instead of many to one is almost entirely useless in that situation as you see people running applications on several different hosts and looking at it all on the screens in front of them.

      This sentence is not only tortured, but ignorant. It's ignorant for a number of reasons, but only one is important: you can run multiple VNC client sessions on a single machine and, if your window management system permits it, tile them so that you can see them all at once.

      Also, what the actual fuck? Are they supposed to be looking at it all, or perhaps part of it, on screens which are behind them? I suppose they could install a mirror above their primary display for this purpose.

      Now, on to the reason why this comment that I am replying to is stupid: I did in fact address your comment in my comment, proving beyond any doubt to anyone with more than two neurons to spark against one another that I did read your blather. I said, and I quote, "Also, there's no reason you couldn't have a single-window vnc-like tool." You went on to complain that I apparently didn't read your comment after I addressed it. If your reading comprehension skills had progressed beyond Hop on Pop then perhaps you would have recognized this fact, and saved me the trouble of chastising you again.

      In summary, take your meds, fucko.

      --
      "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
  4. Lost a good opportunity for a haiku, mate by ifiwereasculptor · · Score: 3, Insightful

    Ubuntu reigns king
    Forces Unity on users
    Its own crown of thorns

  5. problems with multi-monitor support? by Laxori666 · · Score: 3, Interesting

    Why are people still struggling with this? I mean, why is it so technically challenging? It's a simple concept and it's been around for years...

  6. Delay LTS to avoid the risk Mir to become marginal by ceztko · · Score: 2

    I don't have an opinion about Mir and I don't want to express my opinions about last Canonical moves here. What I want to say is that it would be Canonical interest to delay LTS to not risk Mir to become marginal in the years to follow. There would be nothing bad about this move: people that like to stay cut-edge will enjoy a regular 14.04 release. People that need stability will just keep 12.04 another six months.

  7. Mir RIP? by Alomex · · Score: 2

    This might pretty much kill Mir. By the time is released Wayland will likely have taken over and even if Mir is better it will be a case of "too little, too late".