LXDE-Based Lubuntu Will Not Ship Mir Display Server

[mrugiero]

Any closed-source derivative of Ubuntu (or just of Mir) shipped by a carrier on their device should automatically make the device untrusted, as I regard ALL smartphones to be right now. I would prefer that Ubuntu dump the carriers and simply write a known trustworthy OS that users can install on unlocked or rooted devices. Hell, I so distrust all the telecoms that I would insist not only on Linux but also on Coreboot installed myself on any device I was going to use both as a computer and as a modem on a cell carrier's network at the same time. Since the carriers might refuse to connect to such a device, that means a USB port and their USB access stick, or better yet their wireless hotspot and the device's wifi radio, are needed for connection. Nothing else can be trusted by the user and the carrier, as each considers the other malicious.

Then Ubuntu would probably lack funding for Mir. Not that I support Mir, but if they're gonna do it anyway, they better get funding to.

As for distros like LXDE-and older WMs like IceWM, they are meant for older x86 machines and x86 netobooks, both of which slow down badly when using compositing. I will go so far as to say that any machine that either predated Windows VIsta or was considered unable to run Vista when it was sold should be considered a good candidate for a non-compositing, X-only DE. After all, no pre-Vista computer sold with Windoze on it was designed for desktop compositing. Getting rid of compositing really helps both responsiveness and video playback on them.

I agree only partially. Leaving aside the extra memory use, compositing without special effects (just as a meaning to avoid artifacts) shouldn't really add too much overhead, which means there are probably some x86 scenarios where compositing isn't something you need to avoid. Of course, this applies only for the latest generations of x86_32 computers. I wouldn't use compositing on a K6-II with 256MB of RAM and a 16MB Matrox card, of course, but a Pentium 4 with 1 or 2GB and one of the newest AGP GeForce cards would probably work as a charm.

[Luke]

Compositing on Pentium 4 and older machines

I agree only partially. Leaving aside the extra memory use, compositing without special effects (just as a meaning to avoid artifacts) shouldn't really add too much overhead, which means there are probably some x86 scenarios where compositing isn't something you need to avoid. Of course, this applies only for the latest generations of x86_32 computers. I wouldn't use compositing on a K6-II with 256MB of RAM and a 16MB Matrox card, of course, but a Pentium 4 with 1 or 2GB and one of the newest AGP GeForce cards would probably work as a charm.

In fact, Pentium 4's will work fine with old-school compiz, though at a sacrifice of total available capacity if something like playing 720P video is an issue. I found that a 4GHZ Pentium 4 with a Radeon X1600 and the Radeon driver I tested was just able to play video with a totally non-compositing MATE environment (and no Pulseaudio), just short of being able to keep up with compiz-MATE. Almost exactly the same results with my Pine Trail (Intel Atom 1.6 GHZ) netbook, using far newer but small Intel graphics. Back in 2008-2009. with HD video a non-issue, that PIV was run with GNOME 2 Compiz, the performance difference was not noticed. Takes exactly the right load to go just over vs just short I guess. A video card new enough for VDPAU might fix this, but such a card in AGP usually costs too much to justify for a Pentium 4.

On the other hand, most Pentium 3 and older machines you can usually forget compositing. You have to add a (getting hard to find) AGP video card with openGL capability, and only older Compiz will run on the first generation Radeons and similar cards that came in early Pentium 4's and were sometimes dropped into Pentium 3's by owners. A Pentium 3 with enough video card to use XV will play VGA video using MPEG 4 encoding, or DVD video fine but will choke on H264 even in the slightly smaller 360P format. I know this from using a 450MHZ Pentium 3 and an 800MHZ Celeron with several specific videos to benchmark different Ubuntu versions on older hardware a few years ago. Nothing I tested ever outperformed Jaunty Jackalape on those.

There are a few Nvidia GT520's made for a straight PCI slot. I would be curious about how one of these with Nvidia's binary blob would perform on a Pentium 3 "coppermine" with VDPAU and something like 360p H264 and 720p H264. Certainly compositing would work, don't know how bad the CPU load would be to support it. I did once run Compiz with a Geforce 5(something something somthing) card on an AMD Athlon 500 MHZ, it worked but was much snappier running XFCE instead. Firefox bogged down when GNOME 2/Compiz was run. It played VGA video well-but NOT in Flash so video sharing sites were useless on it. This was in 2008, so it was never tested with HD video before I gave it to a friend in need.

[Serge]

It can. A developer's role is to develop solutions *to problems*. As long as Mir doesn't give something Wayland can't give, he has *no reason* to support it, because there's *no problem* to solve.

You make a good point here. Distributions are responsible for integrating software pieces to make functional systems. If KWin runs on Wayland, the onus is on the Ubuntu packagers to make this solution available in Ubuntu. So, in part, I agree with you here.

And favoring a single distro just because, when their solution does about the same thing the solution they were already working on, is a setting incorrect priorities.

Regarding the single distribution solution argument: Well, I think size of user base does matter. It's not something that I personally put a lot of stock in, but it should be kept in mind. More important, I think, is the ocean-full of Ubuntu derivatives. At least several are bound to ship Mir as the default display server.

On the other hand, the possibility it was a business decision doesn't bother me by itself, but the fact they won't admit it (and invent tech facts that get refuted) and that fragmenting the community because of that decision would be a direct contradiction to their speech if it was only business.

I'm the other way around. If a decision causes fragmentation, I can swallow that decision much more easily if it is a technical one (or even an ideological one) rather than a business one. Because fragmentation does cause harm, I'd like to see some benefits brought about by this fragmentation. When it's a business decision, the benefit is a company's profitability, which I have a much harder time accepting than technical or ideological benefits with harfmul side-effects. On the other hand, I too would like it if they called it what it is and not make up technical rationale for a business decision, but call me jaded but I have stopped expecting honesty from for-profit businesses a long time ago. I know it's bad, but I let lying companies off the hook because in my mind that's the norm anyway.

But toolkits lack of support means you can't use most of your software if using vanilla Ubuntu.

My understanding is that there would be Mir:Wayland abstraction layers in the same way that there's XMir and XWayland. Am I incorrect in this?

I'm against the view of business as some entity allowed to do whatever they want. If they make a promise, they should keep it. They said they cared for the community, well, their word should bound them to. And they are not allowed to lie, either. They can invest in whatever they want, as long as they don't promise otherwise, but lying is unnecessary and problematic. If this is a *only business* decision, they should just tell that.

And I'm the other way around. Companies lie. I expect it from them. I think that's the world we live in.

***

And no, Qt is not controlled by a company instead of a community. Ever since open governance Qt is a community project. Digia controls some aspects but not Qt in general.

I was not aware that Qt was under a community governance model. Thank you for the correction.

***

Any closed-source derivative of Ubuntu (or just of Mir) shipped by a carrier on their device should automatically make the device untrusted, as I regard ALL smartphones to be right now. I would prefer that Ubuntu dump the carriers and simply write a known trustworthy OS that users can install on unlocked or rooted devices.

Well, that would be beautiful, but that's not going to happen just yet. Although I am soured by Canonical elsewhere, I actually am excited by the progress Ubuntu Touch is making, because at least I think it's a step in the right direction. That is, all else being equal, I think a world with Ubuntu Touch is better than a world without Ubuntu Touch. Also, I like the interface design as seen in the demo videos.

[mrugiero]

Regarding the single distribution solution argument: Well, I think size of user base does matter. It's not something that I personally put a lot of stock in, but it should be kept in mind. More important, I think, is the ocean-full of Ubuntu derivatives. At least several are bound to ship Mir as the default display server.

I agree on the importance of user base, didn't mean to imply otherwise. But I don't think we have valid statistics to actually determine the users share as I was discussing with (I believe) GreatEmerald, on another thread. Also, I'm not sure how many of them are actually bound to ship Mir by default, aside from Ubuntu flavor (I guess, anyway, the most Ubuntu distro users are also Ubuntu flavor users, but again, this ones are not affected by KDE decisions, since they use Unity).

I'm the other way around. If a decision causes fragmentation, I can swallow that decision much more easily if it is a technical one (or even an ideological one) rather than a business one. Because fragmentation does cause harm, I'd like to see some benefits brought about by this fragmentation. When it's a business decision, the benefit is a company's profitability, which I have a much harder time accepting than technical or ideological benefits with harfmul side-effects. On the other hand, I too would like it if they called it what it is and not make up technical rationale for a business decision, but call me jaded but I have stopped expecting honesty from for-profit businesses a long time ago. I know it's bad, but I let lying companies off the hook because in my mind that's the norm anyway.

I think I expressed it wrong. I am bothered because it fragments, but what I was trying to say is that it bothers me more because of the lying than because of the fragmentation itself. Why? Because I do expect companies who doesn't tell us they care for the community to only care about how they get money. Instead, Canonical spent quite a lot of time telling users they care about free software as a philosophy and what not, to end up fragmenting the desktop. Other than that, I actually agree with you in this subject.

My understanding is that there would be Mir:Wayland abstraction layers in the same way that there's XMir and XWayland. Am I incorrect in this?

The toolkits work as abstraction layers themselves. If there's no support for the toolkits, then I don't know what would they come up with, but yeah, som abstraction layer is one of several possible solutions. But I think they'll go for downstream patches to the toolkits, if upstream doesn't support them. Otherwise, it kind of completely beats the idea of having a different solution, if everything will be run in a Wayland compatible way.

And I'm the other way around. Companies lie. I expect it from them. I think that's the world we live in.

It's reality that they do, of course. But I don't think the "it's their money" argument gives them the right to do so. It's wrong for normal people and it's wrong for companies. And it actually produces more harm this way, since in the case they'd say "we are doing our own because we need more control", probably all the freelancers would be right now off their wagon, keeping fragmentation somewhat controlled.
And at least in my country (Argentina) it's illegal to lie to your customers, and you can start a lawsuit if you think they're doing so. It's called misleading advertising. It's not very enforced, sadly, but the law is there and it should be enforced always. Take it as if it's a verbal contract: they promise something in exchange of your payment (in whatever way you pay), and they must keep their promise.
EDIT: To go a bit further, if we click the 'agree' button on terms and conditions, we are supposed to follow them. The logic behind it is you agreed on those terms to use the software. Well, when you buy a product from a company, the advertising is taken into account, and it should work the same way. They gave you their word. Canonical said they'd always care for the community, they should keep up with it. Those are the terms and conditions they clicked "I agree" to.

[chrisb]

I do expect, however, their carriers to ship modified, closed source derivatives.

How does the Wayland licence stop closed source forks?

[chrisb]

There's also the issue of the license, a lot of FOSS developers refuse to sign over copyright of their code out of fear that the new holder will close source)

How does the Wayland licence prevent closed source versions?

[Awesomeness]

On the other hand, most Pentium 3 and older machines you can usually forget compositing. You have to add a (getting hard to find) AGP video card with openGL capability

Compositing on its own has nothing to do with OpenGL.

How does the Wayland licence prevent closed source versions?

It doesn't but it does create equal ground for all.
The whole Linux graphic stack has historically been under MIT license. Wayland is no different.

Wayland is not about preventing closed source variants, it's about equal rights for everybody. Anyone can take Wayland and do almost anything with it, without the requirement to acquire a commercial license from Canonical.

[chrisb]

It doesn't but it does create equal ground for all.
The whole Linux graphic stack has historically been under MIT license. Wayland is no different.

Wayland is not about preventing closed source variants, it's about equal rights for everybody. Anyone can take Wayland and do almost anything with it, without the requirement to acquire a commercial license from Canonical.

So Wayland is not a solution for all of those FOSS developers who refuse to sign over copyright of their code out of fear that it will be used in closed source.

[mrugiero]

How does the Wayland licence stop closed source forks?

It does not even try to.
I said that in a context where someone said he feared (or something like that) that Canonical would close up Mir. I said what I expect to happen, which is Canonical keeping mainstream Mir open, but allowing their carriers to make closed source derivatives in exchange for their funding.
As I already said several times, I don't think neither the CLA nor the fact it's GPL are inherently bad (well, the GPL is bad for some other POSIX OSes, BSDs for example, but I don't think either Wayland or Mir really thought of them, since both require things that are currently not implemented on those camps). This combination does, however, lead to different rights for Canonical than the rest of the community, being them the only ones able to make/allow closed source derivatives.

Compositing on its own has nothing to do with OpenGL.

It doesn't but it wasn't implied. The rest of the quoted post actually leads to think it doesn't require OpenGL, but mentions cases where it can be done on the CPU without getting bad performance. However, in cases where the CPU is not powerful enough, you need a graphics card, and since backends offloading to the GPU does this through OpenGL, you need the card to support it.

So Wayland is not a solution for all of those FOSS developers who refuse to sign over copyright of their code out of fear that it will be used in closed source.

As I said in a post you already quoted (but taking only the out of context expectation for closed derivatives), this idea comes from a misreading of the CLA. Some people tend to read it like you actually give your copyright to Canonical. IF they are the only copyright owners, there is the chance (never got done, yet) to revoke the GPL or any license it already has. MIT license, as a license, does not allow you to revoke anything. Only the copyright give you this right. That's the point, if you read incorrectly the CLA, you might think they are able to revoke licenses and make it *only* closed source. Big highlight: you don't give your copyright to Canonical, but give them the extra right to sublicense (they are unable to revoke licenses, but *can* make copies with different licenses than the original).

[Awesomeness]

So Wayland is not a solution for all of those FOSS developers who refuse to sign over copyright of their code out of fear that it will be used in closed source.

I doubt those even exist. As I wrote: The whole stack has historically been MIT-licensed. Also Android?s stack is under a similar license.

Mir will be the first Linux display server with meaningful market penetration to ever be under a copyleft license (for certain users at least).
All display server contributors have to come somewhere and while I don?t know the biography of each Mir programmer, the core Wayland developers are Xorg contributors.

However, for outside Mir contributors the situation is totally awkward: They have fewer rights to use the project code than Canonical.
In case of Xorg, Wayland, or Mesa everybody has exactly the same rights. No compromises.

As I said in a post you already quoted (but taking only the out of context expectation for closed derivatives), this idea comes from a misreading of the CLA. Some people tend to read it like you actually give your copyright to Canonical. IF they are the only copyright owners, there is the chance (never got done, yet) to revoke the GPL or any license it already has.

No. The GPL could only be revoked if the code was obtained illegally and the organization never had the right to publish it under GPL in the first place.

Copyright transfer it not possible under EU laws. One can only transfer the right of use. This is why FSF Europe?s ?Fiduciary Licence Agreement? is different from FSF USA?s copyright assignment.
Canonical?s CLA works just like copyright assignment in practical terms: Canonical is exclusively in the position to choose any license for Mir with no legal binding to ever again provide FOSS versions (FSF and Qt Project have these provisions).