Announcement

Collapse
No announcement yet.

Canonical Announces Mir Back-End For Mainline Mesa

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

  • #61
    Originally posted by Siekacz View Post
    I do not want happy-community projects which move on REALLY slowly, because of too much deciding people, too much shared code with other different projects.
    You want sad-community projects instead?

    The hybrid-graphics support is one of the examples - the technology is quite popular for several years, yet we do not have proper support. And it's not because bad corporations dislike Linux, just because Linux can't simply support these things in easy and cheap to manage way. Yes we have emerging Wayland, but it still lacks hybrid graphics support and is in development for 5 years. Way too long. When to expect it being ready? 2016? 2018? Canonical has plans for 2014.
    Ok you have to stop confusing the Wayland standard with its reference implementation. Wayland is a standard, and the wayland standard is ready. The reference implementation is progressing well. Canonical has not enough staff, they have not enough resources, they don't accept community patches, and they don't have the expertise required to create a whole new display server standard AND implementation from scratch.

    This is what doesn't make any sense in your argument: you complain that Wayland is taking too long, so you think it's better for Canonical to start from scratch on an entirely new display server? Why? There is simply no way in hell that Canonical is going to be able to get their display server finished faster than Wayland.

    Putting extra people and money wouldn't help, because of the lack of the control - fighting with community to add some features and remove other would take too much time and too much effort. For example: Canonical wants to throw away system tray and uses indicators to deal with programs in the background. There will be no such thing as system tray in Mir, because it only complicates the whole. Other distributions want tray, just because of "legacy programs". Conflict of goals.
    This is more bullshit. There's no reason why Canonical needs to do their own display server in order to get rid of a system tray. They could use Wayland and do what they want with it. Once again: Wayland is a standard. There's nothing stopping Canonical from maintaining their own Wayland compositor and controlling its development. For that matter, Canonical will have to support legacy programs just the same as any other distro, so your reasoning doesn't hold water at all.

    Comment


    • #62
      It seems to be that it comes as a complete surprise to the community how Canonical developed their NIH syndrome.

      It happened because the "community" is completely impossible to work with in the vast majority of cases. Their work is intermittent and often of low quality, which doesn't necessarily apply to Wayland but Wayland is taking unreasonably long to come out.

      Canonical is a company that cannot afford what they did from 2010-2012. They have duplicated so incredibly much work now. Unity has been written twice, soon thrice, for instance - and they only did so because the GNOME Foundation had no clue what they were doing, and tailspun to copy as much of Unity after Canonical stopped contributing as they could. The Red Hat company said that it was impossible to earn money off of Fedora because it was basically substandard.

      Canonical also constantly gets complaints because of community work - and I suspect it's only a matter of time until they decide to remove Nautilus and GTK+, and obviously X.org is something we all want killed. They've already decided to remove the inconsistent tray icons. The new mouse settings panel is something I complained about because they rip out important features for gamers. Yes I'm ripping quite a lot on GNOME here but it's not just them; they just happened to have made the majority of the software that we found inside Ubuntu. Even Debian, their packaging format, has been forked by Canonical - and that is a very long time ago.

      Right now Canonical has the chance of a lifetime. They have gamers and serious companies knocking on the door asking for an alternative to the restricted webpage that is the future of Windows. Canonical can offer this, but they can't rely on intermittent work and changing goals that don't align with their own. They could have contacted Wayland, but to be honest I think they're fed up with the whole thing, and they're scared of Wayland suddenly changing in a completely new direction that they heavily disagree with, such as what happened with GNOME.

      Now you know why they have NIH syndrome. They've had their own efforts duplicated far too many times, especially by GNOME, to care any more. They still want to provide a free software operating system (onto which you can install proprietary software if you so desire) but they don't want to do it with you. Specifically you. That's why they don't talk to you. They have their own development community now - some of which they've hired, and that's it.
      Last edited by Ishayu; 03-11-2013, 10:00 AM.

      Comment


      • #63
        Originally posted by Ishayu View Post
        It seems to be that it comes as a complete surprise to the community how Canonical developed their NIH syndrome.

        It happened because the "community" is completely impossible to work with in the vast majority of cases. Their work is intermittent and often of low quality, which doesn't necessarily apply to Wayland but Wayland is taking unreasonably long to come out.
        Yeah, like the Linux kernel. The community does such a shitty job on it, Canonical should just fork their own kernel. Right?

        Canonical is a company that cannot afford what they did from 2010-2012. They have duplicated so incredibly much work now. Unity has been written twice, soon thrice, for instance - and they only did so because the GNOME Foundation had no clue what they were doing, and tailspun to copy as much of Unity after Canonical stopped contributing as they could. The Red Hat company said that it was impossible to earn money off of Fedora because it was basically substandard.
        Hey, there's nothing wrong with writing your own desktop environment. As long as it conforms to the standards and is compatible with other desktop environments. You don't like GNOME, go use any of the tons of alternatives - software developed for GNOME works on KDE and vice versa, because we have standards that enable it.

        Canonical also constantly gets complaints because of community work - and I suspect it's only a matter of time until they decide to remove Nautilus and GTK+, and obviously X.org is something we all want killed. They've already decided to remove the inconsistent tray icons. The new mouse settings panel is something I complained about because they rip out important features for gamers. Yes I'm ripping quite a lot on GNOME here but it's not just them; they just happened to have made the majority of the software that we found inside Ubuntu. Even Debian, their packaging format, has been forked by Canonical - and that is a very long time ago.
        The .deb format in Ubuntu is the exact same as in Debian. There's no forking involved there.

        Right now Canonical has the chance of a lifetime. They have gamers and serious companies knocking on the door asking for an alternative to the restricted webpage that is the future of Windows. Canonical can offer this, but they can't rely on intermittent work and changing goals that don't align with their own. They could have contacted Wayland, but to be honest I think they're fed up with the whole thing, and they're scared of Wayland suddenly changing in a completely new direction that they heavily disagree with, such as what happened with GNOME.
        Sigh... how many times do I have to say this. Wayland is a standard. And that standard now has a stable API, so any fearmongering about "changing direction" is totally unfounded. If Canonical wants to develop their own display server, they can do that with Wayland. Nothing in the Wayland standard says that you can't develop your own implementation of it, nothing in it mandates that you have to use Weston. But instead Canonical goes out of their way to introduce incompatibility.

        Let's compare with desktops again. It's fine to have GNOME, KDE, MATE, Xfce, Cinnamon etc. because they all conform to a standard. But if someone went and created a desktop environment that couldn't run any programs written for the other desktops, that'd be an awful idea.

        Now you know why they have NIH syndrome. They've had their own efforts duplicated far too many times, especially by GNOME, to care any more. They still want to provide a free software operating system (onto which you can install proprietary software if you so desire) but they don't want to do it with you. Specifically you. That's why they don't talk to you. They have their own development community now - some of which they've hired, and that's it.
        So basically, you're saying that Canonical devs are assholes. Gotcha.

        Maybe you don't understand what free software is. Free software means collaboration, it's a real-world implementation of the wisdom of the crowds. The idea that the best solution can be found when more brains are working on the problem. It's collaboration by people who use their own products. If this "free software" OS is developed behind closed doors, you have no chance to contribute or have any influence on the development process, everything is dealt to you in a top-down model... how is it practically any different from the proprietary alternatives? Just because it's nominally "open source" doesn't make it free software.

        Comment


        • #64
          Originally posted by dee. View Post
          Yeah, like the Linux kernel. The community does such a shitty job on it, Canonical should just fork their own kernel. Right?
          Linux isn't developed by the free desktop community to the best of my knowledge? I wasn't saying they don't want to work with all communities, I said they don't want to work with THIS community.

          Originally posted by dee. View Post
          Hey, there's nothing wrong with writing your own desktop environment. As long as it conforms to the standards and is compatible with other desktop environments. You don't like GNOME, go use any of the tons of alternatives - software developed for GNOME works on KDE and vice versa, because we have standards that enable it.
          They don't like the standard because it doesn't have compatibility with mobile drivers and likely won't have it. Since they want one display server for both phones and desktops, they decided that it was important for them that you could install Ubuntu Phone on Android Phones, which has a massively larger community, at the cost of the Free Desktop community.

          Originally posted by dee. View Post
          The .deb format in Ubuntu is the exact same as in Debian. There's no forking involved there.
          I was talking more about aptitude and the repository system. I apologize for being unclear.

          Originally posted by dee. View Post
          Sigh... how many times do I have to say this. Wayland is a standard. And that standard now has a stable API, so any fearmongering about "changing direction" is totally unfounded. If Canonical wants to develop their own display server, they can do that with Wayland. Nothing in the Wayland standard says that you can't develop your own implementation of it, nothing in it mandates that you have to use Weston. But instead Canonical goes out of their way to introduce incompatibility.

          Let's compare with desktops again. It's fine to have GNOME, KDE, MATE, Xfce, Cinnamon etc. because they all conform to a standard. But if someone went and created a desktop environment that couldn't run any programs written for the other desktops, that'd be an awful idea.
          Canonical just doesn't care about that - and I don't see why they should. Standards can be changed! Remember OpenGL? This is exactly the fear I was describing - that the standard changes.

          While that change was something people got over and actually came to appreciate, it was a change that deprecated a lot of stuff and riled people up pretty bad and a lot of software companies had to rewrite their renderer - such as Blizzard for instance. (To this day it still has 2 different OpenGL renderers! )



          Originally posted by dee. View Post
          So basically, you're saying that Canonical devs are assholes. Gotcha.
          No I didn't say that at all. I just said they don't want to work with you.

          Originally posted by dee. View Post
          Maybe you don't understand what free software is. Free software means collaboration, it's a real-world implementation of the wisdom of the crowds. The idea that the best solution can be found when more brains are working on the problem. It's collaboration by people who use their own products. If this "free software" OS is developed behind closed doors, you have no chance to contribute or have any influence on the development process, everything is dealt to you in a top-down model... how is it practically any different from the proprietary alternatives? Just because it's nominally "open source" doesn't make it free software.
          I do perfectly well understand what free software is. It seems to me that you are attributing things to free software which it isn't. Free software is not a community effort by definition. The only thing it means is that you give the source code free of charge, give the rights to change (aka fork) the code and give the rights to redistribute the source code in full including your changes if you so desire. It has NOTHING WHATSOEVER to do with the Free Desktop community having a final say and it particularly has nothing to do with standards - in fact it is diametrically opposed to the idea. (I still think we need them though) It doesn't mean you have to develop everything out in the open from the very beginning either.

          I'm not outright defending Canonical. I do think they should have tried to pull Wayland in their direction or just forked Wayland instead of starting from scratch. I was simply trying to explain to you why this happened. EDIT: Clarifying - it is because the last time they tried to work with the free desktop community, in that case specifically the GNOME Foundation, they got ZILCH out of it.
          Last edited by Ishayu; 03-11-2013, 10:47 AM.

          Comment


          • #65
            Originally posted by Siekacz View Post
            OS X is getting outdated and stinky. Ubuntu is the only hope to make personal computiing powerful and easy. Addition of being OpenSource is nice, but it is not enough for me.
            What has Canonical said that makes you think that they will be one of the few people in the history of ever to make an operating system you like?

            Comment


            • #66
              Originally posted by Siekacz View Post
              I do not want happy-community projects which move on REALLY slowly, because of too much deciding people, too much shared code with other different projects. The hybrid-graphics support is one of the examples - the technology is quite popular for several years, yet we do not have proper support. And it's not because bad corporations dislike Linux, just because Linux can't simply support these things in easy and cheap to manage way.
              Reasons are documentations, non-disclosure agreement and software patents. However, hybrids like AMD E-series work out of box although they are not optimized for heavy 3D videogames like Crisis 3.

              [quote] Yes we have emerging Wayland, but it still lacks hybrid graphics support and is in development for 5 years. Way too long. When to expect it being ready? 2016? 2018? Canonical has plans for 2014.
              Putting extra people and money wouldn't help, because of the lack of the control - fighting with community to add some features and remove other would take too much time and too much effort.
              Canonical has not even once participated to Wayland development, it simply waited for other distributions to work on heavy lifting, pick up the result with some modifications and claimed credits for themselves and add CLA when needed.

              For example: Canonical wants to throw away system tray and uses indicators to deal with programs in the background. There will be no such thing as system tray in Mir, because it only complicates the whole. Other distributions want tray, just because of "legacy programs". Conflict of goals.
              Problems are those modifications are horrible hacks including Unity itself. Mir is only a trail paper with developers lacking clue about Wayland hence the wiki which was modified since then. Realistically, Mir will not take off because the project has the burden to complete its development in short time.

              I want well designed, beautiful, OPEN product with no compromises. That's what Ubuntu's going to be.
              Canonical Ubuntu will never be what you are looking for, from the outside it might look beautiful but the inside is rotten.

              Comment


              • #67
                Originally posted by Ishayu View Post
                Linux isn't developed by the free desktop community to the best of my knowledge? I wasn't saying they don't want to work with all communities, I said they don't want to work with THIS community.
                But Canonical IS unwilling to work with communities.

                They don't like the standard because it doesn't have compatibility with mobile drivers and likely won't have it. Since they want one display server for both phones and desktops, they decided that it was important for them that you could install Ubuntu Phone on Android Phones, which has a massively larger community, at the cost of the Free Desktop community.
                Bullshit, there's nothing in the Wayland standard that prevents it being used with mobile drivers. In fact it has already been done as a proof-of-concept.

                I was talking more about aptitude and the repository system. I apologize for being unclear.
                Aptitude (and apt which it is a front-end for) is also a part of Debian and developed by Debian. Nothing Ubuntu-specific there either. PPA's, ok, that's one thing that's Ubuntu-specific, but repositories in general are not.

                Canonical just doesn't care about that - and I don't see why they should. Standards can be changed! Remember OpenGL? This is exactly the fear I was describing - that the standard changes.
                Bullshit FUD. Wayland API has been declared stable and there's no reason to assume Wayland would change their standard, other than BS fearmongering and propaganda. Even if the standard did change (unlikely - has Xorg ever made such changes in the standard?) Canonical could just choose to ignore the changes.

                And OpenGL is totally irrelevant here.

                No I didn't say that at all. I just said they don't want to work with you.
                It doesn't really matter if Canonical wants to work with me or not because I don't work on display servers.

                I do perfectly well understand what free software is. It seems to me that you are attributing things to free software which it isn't. Free software is not a community effort by definition. The only thing it means is that you give the source code free of charge, give the rights to change (aka fork) the code and give the rights to redistribute the source code in full including your changes if you so desire. It has NOTHING WHATSOEVER to do with the Free Desktop community having a final say and it particularly has nothing to do with standards - in fact it is diametrically opposed to the idea. (I still think we need them though) It doesn't mean you have to develop everything out in the open from the very beginning either.
                Who has said anything about FreeDesktop community having a final say? Free software is opposed to standards? Wow, what? Where do you get such preposterous ideas?

                No, if you're developing something that you hope to become a standard for the community, it should be developed in the open and with collaboration, not with some kind of skunkworks team behind closed doors.

                I'm not outright defending Canonical. I do think they should have tried to pull Wayland in their direction or just forked Wayland instead of starting from scratch.
                I was simply trying to explain to you why this happened. EDIT: Clarifying - it is because the last time they tried to work with the free desktop community, in that case specifically the GNOME Foundation, they got ZILCH out of it.
                GNOME has nothing to do with Wayland.

                Comment


                • #68
                  Originally posted by dee. View Post
                  No, if you're developing something that you hope to become a standard for the community, it should be developed in the open and with collaboration, not with some kind of skunkworks team behind closed doors.
                  I think that is the real interesting thing here - the fact that many of the people decrying standards want us to use their's instead.

                  Comment


                  • #69
                    This post explains everything. They want display server to be in developed in TDD style. Wayland is not TDD. So: reimplementing a lot of the stack + writing tests for existing code. That's an overkill and they can't afford bending Wayland to their needs.

                    Comment


                    • #70
                      Originally posted by Siekacz View Post
                      This post explains everything. They want display server to be in developed in TDD style. Wayland is not TDD. So: reimplementing a lot of the stack + writing tests for existing code. That's an overkill and they can't afford bending Wayland to their needs.
                      Don't be stupid.

                      Writing tests for Wayland would be a lot less work than creating an entirely new display server AND writing tests for it.

                      Mir was created so that Ubuntu had control over the project, copyright and otherwise. It's the only explanation that makes any sense.

                      Comment


                      • #71
                        Originally posted by Siekacz View Post
                        This is best picture of so-called "FOSS Community". Hate, jealous, hypocrisy and IBC (Invented By Canonical) syndrome.
                        What a lie. What most people care about is the FOSS to be developed in the OPEN on terms EQUAL to all parties and COMPATIBLE to most distrobutions. Mir offers nothing but CLAed skunkwork which have a HUGE hidden cost. Waylands reference compositor and the work at GTK/shell is heading towards CSD. This is quite orthogonal to Mirs directions. So even by accepting the CLA and Mir not being distro agnostic you have a future waste of engineering and development hours much larger than the hours wasted at Mir so far.

                        The only sane choice is to stop Mir now and turn back. Instead of doing this Canonical have started lying and defaming the better choice. THATS what pisses many people off.

                        Comment


                        • #72
                          I smell lack of knowledge about TDD. The purpose of Test-Driven Development is to write tests BEFORE writing the program. If you write a program before tests, it's not TDD.

                          "Maybe we want Wayland, but we'll need to write an input stack, patch the Mesa EGL platform, and redo the WM handling in all the toolkits." - isn't that enough?
                          Last edited by Siekacz; 03-12-2013, 06:24 AM.

                          Comment


                          • #73
                            Originally posted by dee. View Post
                            But Canonical IS unwilling to work with communities.
                            Clearly. This is why they're reaching out to their community to help them develop ALMOST ALL of the apps that are now supposed to replace the GNOME apps (I called it, not it's a news item on this site. More surprises to follow >_> ) or how they're willing to fund what can only be described as forks of Ubuntu. (Xubuntu, Ubuntu GNOME remix, Edubuntu, etc.)

                            Originally posted by dee. View Post
                            Bullshit, there's nothing in the Wayland standard that prevents it being used with mobile drivers. In fact it has already been done as a proof-of-concept.
                            There is nothing preventing it from being used except the fact that it can't be used because the standard is incompatible with the vast majority of Android display drivers out there. Good luck telling all the phone companies to rewrite their driver stack for all the phones already in circulation LOL.

                            Originally posted by dee. View Post
                            Aptitude (and apt which it is a front-end for) is also a part of Debian and developed by Debian. Nothing Ubuntu-specific there either. PPA's, ok, that's one thing that's Ubuntu-specific, but repositories in general are not.
                            They work slightly differently in Ubuntu. This is merely a tangent anyway. Nevermind.

                            Originally posted by dee. View Post
                            Bullshit FUD. Wayland API has been declared stable and there's no reason to assume Wayland would change their standard, other than BS fearmongering and propaganda. Even if the standard did change (unlikely - has Xorg ever made such changes in the standard?) Canonical could just choose to ignore the changes.

                            And OpenGL is totally irrelevant here.
                            OpenGL itself is somewhat irrelevant, but the point that I'm making is that just because you have a standard does not mean it can't suddenly change completely. Yes, you can ignore it, but that causes people to be stuck with busted old broken software. OpenGL would be a clear example of this. Calling it "bullshit FUD" is just plain ridiculous.

                            Also, unwillingness to ignore the standard is what has led to the X.org fiasco, just FYI. But of course, if everyone ignores the standard, there is no standard, so at that point it all becomes a bit moot, yes?

                            Also - I don't know if you're noticed this - but Mir is basically Wayland with extensions and a different implementation. It isn't terribly different and it already runs Qt just because Qt was ported to Wayland, for instance. This is why Canonical can so safely say that e.g. KDE will run on Mir.

                            Originally posted by dee. View Post
                            It doesn't really matter if Canonical wants to work with me or not because I don't work on display servers.
                            Maybe you ought to learn some grammar. "You" can also refer to many different people - or indeed several communities. I don't care whether you, as in you alone specifically mr. Dee., aren't a display server engineer. That was not what I was saying.

                            Originally posted by dee. View Post
                            Who has said anything about FreeDesktop community having a final say? Free software is opposed to standards? Wow, what? Where do you get such preposterous ideas?

                            No, if you're developing something that you hope to become a standard for the community, it should be developed in the open and with collaboration, not with some kind of skunkworks team behind closed doors.
                            Free software has, as its core value, that you should be able to add and remove any part of a program you wish in any way you wish.

                            If part of the program is part of a standard you don't want to have due to whatever reason, we've got a problem with standardization vs. free software. You have put restrictions on what your community should and should not do.

                            And I repeat: Free software is supposed to benefit the community, yes, but that doesn't mean the whole community needs to develop it.

                            In practical terms I agree, which I stated, and I'm not saying Canonical has the better idea here. I'm just simply telling you to stop trying to persuade me that free software has to be developed openly. Free software is a publication ideal, not a development ideal.

                            Originally posted by dee. View Post
                            GNOME has nothing to do with Wayland.
                            They often work together and several members work in both at once.

                            But hey, it's not as if Canonical isn't sick and tired of the X.org guys, for instance, either. Revving X.org keeps breaking Ubuntu computers and Canonical is quite possibly a little tired of having complaints about it.
                            Last edited by Ishayu; 03-12-2013, 02:28 PM.

                            Comment


                            • #74
                              Originally posted by Siekacz View Post
                              I smell lack of knowledge about TDD. The purpose of Test-Driven Development is to write tests BEFORE writing the program. If you write a program before tests, it's not TDD.

                              "Maybe we want Wayland, but we'll need to write an input stack, patch the Mesa EGL platform, and redo the WM handling in all the toolkits." - isn't that enough?
                              Wayland is just a protocol, it's kind of tough to do TDD with that. The test come in when you write a compositor for it - such as Weston. Or Kwin/whatever Gnome is doing.

                              Everyone assumed Ubuntu would write their own Unity compositor, whether it was called Mir or just part of Unity. That's where the TDD comes in. No one would have had a problem with that.

                              It's true enough that there might have been small bits that were developed before tests, but you can write tests after the fact for those occasions. Blindly following TDD without using any common sense is worse than having no tests at all.


                              Anyway, let's go over that list you linked earlier, shall we?

                              1. Yes, X sucks. Everyone agrees.
                              2. Sure, Weston isn't for you. It's probably not for 99% of people. That makes perfect sense. It's why Unity should have had it's own Wayland compositor.
                              3. Ok, so 9 months ago there was no input stack for Wayland. There is now, though. And now Mir doesn't have an input stack. So, basically this reason is reversed now - it's a reason to use Wayland instead of Mir.
                              4. Server-side allocations. Yeah, ARM GPU's suck and have special requirements. Requirements that wayland easily provides. A dev already showed you, just override 2 methods in your compositor and you are good to go, and it's already been tested and proves to work. Easy. No reason to write an entirely new spec for this. And frankly, the fact that none of the Mir devs knew this is one of the big red flags i have. How can we have confidence in these guys when they can't even see the obvious stuff like this?
                              5. This point basically boils down to them not liking the Wayland API and wanting to replace it, for no particular reason that they explained. I think basically they just want control to do whatever it is they want, without having to get approval and reviews from upstream.

                              Which basically boils back down to my previous post - they want control of the project, copyright and otherwise, without having to deal with the community.

                              I really wish they'd just come out and admit that. It would have gone a lot better for them if they were honest about that point rather than trying to throw together all these half-baked justifications that don't make any sense and just make them look incompetent.

                              Comment


                              • #75
                                At this point, it looks like we want something like Wayland, but different in almost all the details. It's not clear that starting with Wayland will save us all that much effort, so the upsides of doing our own thing - we can do exactly and only what we want, we can build an easily-testable codebase, we can use our own infrastructure, we don't have an additional layer of upstream review - look like they'll outweigh the costs of having to duplicate effort. Therefore, Mir.
                                They Admitted that.

                                Comment

                                Working...
                                X