Canonical Posts 15 Mesa Patches To Support Mir

[jakubo]

for me it works on nouveau (Nvidia 8600) radeon RV535 and intel GMA945. please dont judge on your ONE experience... post a bug-report instead
there is just one thing that kills the CPU completely (performance wise): moving mouse pointer over a flash video area!

[mrugiero]

Whatever happened to Linux being about choice? If a company wants to create their own display server, what's wrong with that? We should start putting restrictions on what display server you can use in Linux? That surely sounds pretty restrictive.

Mir and Wayland are both the future.

Should we start hating on Google now for not using Xorg or Wayland for Android and instead making their own?

If a company wants to create their own display server, they can. If a project doesn't want to accept that company's patches, they can.
They can support themselves out of tree if they want.
It doesn't make it any better. Nobody is banning them from doing whatever they are pleased to, but we are free to be upset about this choice.

Google doesn't fragment the desktop, that's why most don't care a little bit about them, they are on phones and tablets and nobody targets Android for the desktop.

While it's true, that Canonical started all this, the same could be said vice versa about the Linux Community ever since Ubuntu started. If we really get fragmentation it's the fault of both not being able to work with each other, but instead blaming and rejecting (although most of the blame and rejections don't come from Canonical).

No, the same couldn't, because working with each other isn't the same as dropping all your previous work because of someone who had a whim. Working together is easier if everyone share their needs. Canonical forgot to do that and instead just assumed that what was lacking on Wayland back then would lack forever, without even asking for what they needed. If it were that they asked Wayland devs for the things they needed and they told them to go play elsewhere, then you'd be right. But not wanting to work with them now on the Mir thing is not being unable to work together, is not wanting to waste efforts doing again, from zero, all the work they already invested on Wayland support.

Android is developed behind closed doors too...

Do you see anyone praising Google around here? We just don't name them constantly because Android doesn't affect us, desktop users. Can't affect where your market share is and will always be zero. What Ubuntu does, does affect desktop users, because is one of the most used on desktop and supposed to be for desktops too.

Yes, that's what I meant, when I said that Canonical started all this. That's not only true for Mir. My point is, that we'd benefit from Canonical being tamed by the Linux Community instead of being ignored (or even fought). The latter is, what may actually lead to fragmentation. From a moral or ethical perspective, the Linux Community may be on the better site, but that alone won't help anyone, if it'll be Ubuntu where the party is at. But maybe Canonical will eventually fail and things will calm down again - but than, I'm afraid, Linux will develop backwards into a niche and we'll again need to rely on Apple or Microsoft, if we want to do anything more than compiling our own kernels.

This has been discussed to death, but for anyone outside of Ubuntu benefiting from Mir, we need at least a compromise of a stable API. No other desktop aside of Unity would be able to follow this (possibly) constantly breaking API. And, being there seems to be no extra benefit on using Mir compared to using Wayland and that a lot of work was already put on supporting the latter, it makes no sense. It takes more work for no benefit, so what's the point? Also, part of Linux being about choice is that not everyone has to follow a single company's orders. And since they are making bad decisions, I'd have a hard time to follow them. For example, XMir on 13.10 and 14.04 is a decision that I think will directly harm users, so why should I go along with that?

After getting rejected pretty much everytime they did something, this isn't a huge surprise.

They could have been rejected a thousand times on other areas, and maybe wrongly, but that doesn't make it any less wrong to reject them on this particular area. Particularly, the whole Mir thing is the only one I actively opposed to. Then, there was Unity, which didn't fit me, but I don't think that hurts anyone. upstart is a good init system, IMO. Mir still has no real, explained, reason to exist, other than Canonical saying "hey, remember when we said we cared? well... we kinda lied, but hey, look, SHINY!".

That is naive point of view.

How many FLOSS projects do you participate in?

Its NORM that any company who wish to build something ONTOP of given project, may at some point join in on development on supporting projects. That what Canonical is doing here. JOINING Mesa developemtn effort. In order to further their development of Mir.

There is nowhere here any message about abandoning any patches.


And before you replay more. Mesa code repo is OPEN.

Go find any code Canonical added, and then orphaned (but still relied on it).

Yes, and most of them actually commit things that aren't just useful for them. The time being, the only things they are committing are to support themselves, and bring no benefit to anyone else. Considering this, it's not a big deal if they get rejected. Will anyone suffer? Nope, they can support themselves out of tree like they'll do in 13.10. If they accept them, there is a chance to be orphaned. Also, go find a project not created by them who gets more than one commit.

Apparently, you do not know what you are talking about (historically)... Ubuntu was considered a "hostile fork" by it's upstream Debian developers at the time of it's inception. Shuttleworth and friends decided rather than improving Debian to be more user-friendly, they would instead just fork it... So no, it can't be "said vice versa". ~ You also have to remember things like Shuttleworth threatening other distros to rebase on Ubuntu (over debian) or "die" (Mepis)... or the conscious choice to break their toolchain (from debian), etc...

I don't agree here. About the threats, I have no idea, so I will not make any comment, but I ask you to explain more deeply so I can understand the situation.
But about the fork? First, distros, except for their package managers (which is the same as in Debian, so doesn't change anything) doesn't add real fragmentation: you don't need to use another toolkit or anything like that, the same software can easily run on different distros. And considering Ubuntu was mostly about making Linux marketable, it kind of makes sense a clean start. I'm much more concerned by their coding than by their forking distros. I don't quite know how did audio get fragmented aside from OSS and ALSA, so I have no idea if Canonical is to blame there. On the init systems situation, they made upstart first, so I think it would make more sense to blame whoever did systemd (Red Hat?). Even though, the init systems doesn't affect much.

Alex, you obviously do not know your history and do not have a clue about this stuff, to be making such silly comments ~ anyone who was around at the time of Ubuntu's inception knows what you are saying is utter BS. you are trying to twist history to make it seem as though the community caused all of these issues, when in reality ~ you could not be further from the truth...

With this, I agree. Blaming the ones who made a solution first for the fragmentation is kind of nonsense. Is not like it was just sitting there doing nothing. Most of the ecosystem already had a lot of work invested in working over Wayland by the time of Mir's announcement.

MIR's licensing (you have to assign Canonical your copyright) basically makes it impossible for any other distribution to use MIR without giving up control over a very important part of a desktop linux system.
Licensing is one of the main concerns regarding MIR. So in this case, Canonical actually deserves the blame

Nah. For a start, you can not "assign your copyright", because in some countries that's just illegal. You give them the right to sublicense, which basically is what everyone can do with Wayland's license. In any case, the problem is that it doesn't give everyone the same rights. But as of control of upstream (on the license level, they will still have the last word on accepting and rejecting patches on the main repo, being their repo) it isn't any worse than Wayland. Of course, a lot of people, including myself, wouldn't commit their work to a project that gives a very specific group of the project more rights than everyone else, independently if the general case is free software; I believe all developers should have the same rights over the code. All can sublicense (e.g., MIT license) or nobody can (e.g., GPL without CLA).
What I wonder is how this situation works regarding forks. Is the CLA still valid for forks? Or it's just on upstream?

1) Canonical will be successful, but be fragmented away from the rest of the Linux world.

2) Canonical will fail and Linux (for endusers) will continue to be niche, thus endusers will still be forced to rely on MS or Apple.

3) Mir alone will fail thanks to their too early, PR lead, introduction as XMir, and Canonical will try and fix the harm done going Wayland.

I sincerely hope Canonical at least open their eyes and see XMir is a bad idea.

Anyway, I explained previously why the path you suggest leads to less fragmentation is flawed.

I don't see Ubuntu as an hostile fork. That's stupidity from the community.

If people already in charge of a distro does not share your vision of how a usable and easier for the user a distro should be, then you can only do a fork.

Speaking of Wayland vs. Mir, I do not like very much what they're doing. They could have made a fork of Wayland like they did with Ubuntu and Debian (which to a large degree their packages are identical and all the source code from Canonical is available) and add or modify whatever they wanted and just try to keep the patches up to date if it's possible, in case Wayland's source code changed continuously and broke every time and recompile, well, it would be of course much more cumbersome.

On the first two comments, I completely agree. Also, I don't think distros does any harm.

About the last one, for a start, forking should be always the last resort. Before even forking, they should check they don't agree, not just assume that like they did. IF they didn't share their views, then they might have chose to fork, and try to keep API compatibility, so no fragmentation is added. But even so, all they do with Mir seems to be doable as a Wayland compositor. Server structure: check, you can make your compositor a display server, even though Wayland doesn't impose this architecture. Server side allocations: the same, you can make your compositor do all the allocations. This would probably break compatibility with apps that use a client side allocation model, but it would probably be not too hard to fix, contrary to a whole different API as it is right now. All of the other features I read about are actually copied from Wayland (i.e., using libhybris, etc.). MAYBE utouch isn't copied, but since Mark said at some point their utouch could have been useful for Wayland, I can only assume by their own words it could be easily used within a Wayland compositor.

Nope the Linux Community helpped Ubuntu and mark used his cash to pay off developers to get them using Ubuntu and his cash for PR and more cash to say hey come Develop Ubuntu see all the green now we have this out of tree thing called Ubuntu and some new Vaperware called Mir

do you work for Canonical Ubuntu is one of the most lame Linux's for the endusers? you want Tumbleweed or Arch Linux for End Users

Nobody denies that. But out of enterprise users, Ubuntu wins on popularity. That's what defines it to be niche or not niche, the popularity. The things you named just defines the software quality. And since for having commercial apps, market share (i.e. popularity) is more important than quality, the desktop would probably suffer on that area, specifically in games support. On the other camps, we will probably see no difference if Canonical just pops. I'd name hardware support and professional applications, but enterprise users do count for that, so no change there either.

Did you checked ubuntu 13.10 on radeon/nouveau/intel drivers recently?

I did. Even when the gross corruption the pic he constantly puts is not anymore, there are some problems that make the experience quite shitty.

I hope the Mesa team is more mature than you are. You sound like an angry little kid.

Even though I have no special hope on either side (and kind of want Mir to get good support or to fail miserably, but nothing in between, since in between still means fragmentation, at least we should have a good product causing it), most Canonical defenders sound really childish (well, maybe only BO$$, but he's so persistent he's the one who comes to mind), and stating a hope doesn't sound childish to me.

#2 - I do not believe in this scenario. Canonical != Desktop Linux and people need to stop pretending that it is, as "the one and only" taker - it's not. I think you will find that as time progresses, that upstream(s) will continue to improve the stack and there will come to boiling point where most ease of use / user-friendly distros are very competitive with each other (and Ubuntu) and provide a nice OOTB experience for end-users. - the big "end-user" advantages of Ubuntu (as some see them) have slowly been disappearing, as upstream stacks get better.

I don't think either the support or the ease of use is Linux real problem now, but PR. And that's what Canonical is good for, PR.
You can have 30 perfectly supported, easy as hell to use distros, that if newbies don't know which of them "is Linux" will have at most 4% market share between all of them. We already have an easy to use system in almost all of the general use distros. I heard GREAT things about Fedora. But I heard nothing about it outside of people who already uses Linux, even when it's easy to use. Right now, a lot of distros are even easier to use than Windows in several areas. But again, Ubuntu is THE popular one, the one people who doesn't yet use Linux know about. There could arise other, anyway, but I wanted to point out that PR is our current problem, and that it affects commercial support, and mostly nothing else. Anyway, I use mostly free software on Linux (and most of my time in general), so it doesn't really change my life to get commercial apps support here.

you are stupid guy, how many people get patches to mesa better, even microsoft. we never know if is important or not for others, itś open source i can t see the problem.

ppl like you f**** linux universe.

liberty of choice and we use what we want

Let's think. First, I don't know about any MS code on mesa. I know of it on the kernel, for Hyper-V support.

On "people using it", let's consider a few things:

- Officially, Mir is only for Ubuntu, at least right now. This means you either use Ubuntu, or you have to build it yourself.
- If you build the software, you probably know, and is not too hard for you to do so, how to patch mesa locally, and it's likely that the out of tree patches will be in the same place the sources are.

So, the only case where it matters if it gets merged is for distros using it, since the idea is the distro user doesn't need to build Mir. Which currently are only Ubuntu. So, Ubuntu will patch locally as they are doing. Does it change anyone's life? No. Does accepting the patches create a possibility for Canonical to orphan them and mesa having to support it? Yes. I don't think it's likely, but if they choose to reject it because of that, I'd accept that reason, because that possibility is real. Anyway, as long as they don't break things for everyone else, I'm not against mesa accepting the patches, and I know mesa wouldn't accept them otherwise. But they have no obligation to do so, either moral or legal.

yeah see here



thats Mir on non Intel Opensource Drivers

That was Mir on non Intel OSS Drivers at least three weeks ago. It (kind of, in a crappy way) works with my Radeon card.

Wow there is way too much hate going on here. I literally laughed when I read someone suggesting Arch for regular users who just want their computer to work (I do love Arch, btw.!).

I didn't laugh, but it sounded really weird. I can imagine my friends having to deal with manually defining what they want their OS to have, when they hardly know what a driver is.

Ubu 13.10 up to date? (Fresh instal preferably)

Ubu 13.10, up to date, fresh install, does not include Mir. You still need to use a PPA. I guess it will be there by default before the betas.

[blackout23]

Ubu 13.10 up to date? (Fresh instal preferably)

Mir broken like hell at the moment if you'd look past the PR videos they made for a moment.

https://plus.google.com/108668411027851722546/posts/jKSkoxeZeGN

[Vistaus]

I hope the Mesa team rejects the patches.

Canonical gave a big fuck you to the Linux community for going with their own inhouse display server which causes fragmentation.

It's crappers like you who are the fools. Many Linuxers (which category you seem to fall in) are like "NOOOO CANONICAL DOESN'T DO UPSTREAM STUFF, FUCK THEM" and (apart from the XWayland patches) now they finally contribute something upstream and they get the middle finger from you. WTF dude. Appreciate the fact that they finally do something upstream.

[synaptix]

Shall I mention this again?

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

[LinuxGamer]

I didn't fight that. However I don't think, that "hostile" is an adequate classification for forks in general, but whatever - let's just leave it that way.

Edit: Whoops, just noticed that you weren't referring to me, but I'll still leave my thoughts on that.



Technically you're right. However, humans are not machines and they make decisions based on attitudes. Considering Ubuntu/Canonical "hostile" is an attitude, that will lead to decisions, which will further fragment Linux.

And you're also right, that it is within the rights of the community not to accept patches - as I said, morally the community has the lead. But if no one accepts Canonicals patches, will that decrease or increase fragmentation? Canonical started it by actually patching things (you have to admit, that in order to do, what they have in mind (e.g. Unity), this is barely avoidable), but the community could have minimized damage by integrating these patches/ improving them together with Canonical. I know, this seems to be asked for far too much, but in the end, that would have been more useful, wouldn't it?



I know this. I still favor this over using software from MS or Apple. Sure, I could use other distros (which I do from time to time), but in the long term, that doesn't seem to work out.

Furthermore, I don't think that isolating themselves is an actual goal of Canonical. They just don't seem to bother that much.



Canonical's big contribution to that is barely of a technical nature, it's marketing. How many vendors are actually selling machines with various distros pre-installed (that means, not only Ubuntu)? There is ZaReason - and well, that's all I heard of. System76 is (afaik) only selling Ubuntu. And both of them are pretty small. On the other hand, e.g. HP, Dell or Asus (at least with Note/Netbooks) are selling some machines with Ubuntu pre-installed - only Ubuntu (in terms of Linux support).

As I said, Canonical is not the only one capable of pushing Linux to the masses, but right now they're the only one doing so. Just having great tech lying around somewhere is not enough, you also need to get it ~sold~. If it was only about having good code, why would MS dominate the desktop?

Or let's just put it this way. If Canonical fails, who would then bring Linux to the masses? I'm not asking for someone who could do this, but who actually would. Valve seems to have some interest in this, however I wouldn't bet on them taking over the desktop os business for Linux.

desktop os business for Linux LOL ROFL LOL some more Stop it BO$$ Debian has been on Desktop Linux For Many Many Years we dont even know how many desktop users Ubuntu has as many keep moving away from it and the Real desktop os business for Linux is suse and RHEL any ways so Ubuntu can come to a EOL and no one will care not one bit if i remeber you can get even get Arch Pre installed and as a site admin seen more non Ubuntu user then Ubuntu users
Btw Canonicals patches most of the time Suck and what if they break User Space? for non Ubuntu DE's?

[alexThunder]

mrugiero, you should have sticked to the latest state of the discussion, instead of commenting on points, which already have been developed further.

Still there's one point I'd like to adress:

we need at least a compromise of a stable API

https://lists.ubuntu.com/archives/ubuntu-devel/2013-June/037342.html

>> - When will MIR have a stable API/ABI?
>
>
> The plan is for libmirserver to have a stable API/ABI by the time we release
> Unity 8 (again, around the 13.10 timeframe). We are stabilising libmirclient
> at the moment since it has more consumers than the server API. Though we
> would expect more functionality to be added to both APIs post 13.10.

LinuxGamer:

The big companies, who sell machines runnung linux - which distro do they ship? Does Dell advertise RHEL, HP Arch or Asus Debian for their enduser machines? No? What would you guess, why they're not?

[LLStarks]

An ilo driver patch...

How edgy.

I thought Intel Gallium was just a joke project.

[synaptix]

The big companies, who sell machines runnung linux - which distro do they ship? Does Dell advertise RHEL, HP Arch or Asus Debian for their enduser machines? No? What would you guess, why they're not?

Canonical is the Microsoft of the Linux world. Practically same tactics.

Also if I were to get something from say System76, I'd have them ship me the computer with the drive blank and not with Ubuntu on it.

[johnc]

Google doesn't fragment the desktop, that's why most don't care a little bit about them, they are on phones and tablets and nobody targets Android for the desktop.

ChromeOS is a Linux desktop. And they're probably selling more Chromebooks than all other Linux desktop / laptop sales combined.

Do you see anyone praising Google around here? We just don't name them constantly because Android doesn't affect us, desktop users. Can't affect where your market share is and will always be zero. What Ubuntu does, does affect desktop users, because is one of the most used on desktop and supposed to be for desktops too.

Quite the contrary... Android is heralded as this great example of how successful and dominant Linux is in the consumer space. The point is... if you want success in the consumer market, you have to do things a bit differently than what the "Linux community" wants. So all this piling on Canonical is just hypocrisy.