XMir postponed in Ubuntu

https://lists.ubuntu.com/archives/ubuntu-devel/2013-October/037695.html

by ovis • 13 years ago

117 110 13 years ago

lists.ubuntu.com

comex2• 13 years ago

Does Canonical have any coherent explanation for why they are still pressing on with their own incompatible rewrite of a display server when a perfectly good one was already in development?

acabal• 13 years ago

My guess is that they're doing it so they can have control over the pace of development and features. I think Shuttleworth got burned by the early-days Gnome Shell debacle--he felt that they had no input on or control over what Gnome Shell was to become, so he went his own way with Unity.

Mir might be his attempt to control the stack in anticipation of getting burned again. After all, supporting Wayland means working in a committee-ish environment, with principal developers not on the payroll and thus not able to be whipped at a certain pace, and without a dictator's say in the direction of the product. What if he wants feature Z included in the next dev cycle to support cool new mobile hardware? With Wayland, he'd have to play politics with the developers and pray. With Mir, he can just order it.

Not to say that I think it's a good decision. In fact I think it's one of the most ruinous decisions for the Linux community that Canonical could have possibly made. Graphics support has been the biggest thorn in Linux's side for years, and dividing the ecosystem at a time when Valve is poised to finally revolutionize things is pretty despicable. But I can see the short-sighted business logic behind it.

Jasper_2• 13 years ago

I think Shuttleworth got burned by the early-days Gnome Shell debacle--he felt that they had no input on or control over what Gnome Shell was to become, so he went his own way with Unity.

As someone who was there from the beginning, this is false. The hackfest where some of the initial goals of GNOME Shell were planned out in 2008 at Canonical offices. We were wondering why some of the Canonical guys who were listed as RSVPing weren't there... until we went to lunch and saw that they were having their own private meeting. I don't think we'll ever know what was discussed -- perhaps their indicator system.

Canonical employees were participating in the Boston hackfests in 2008 and 2009 and we welcomed their presence; they actually had a lot of good ideas which we integrated. I, personally, never had any issues with Canonical's work and I think all of my coworkers would agree. I'm really not sure where this perceived "us vs. them" animosity came from, but I never felt it personally.

They're still absolutely welcome to contribute back to GNOME if they want to (and, to be fair, some still do).

epsylon2• 13 years ago

Valve is poised to finally revolutionize things is pretty despicable.

Valve's announcement is only a few days old. Mir precedes it by months.

StavrosK• 13 years ago

After all, supporting Wayland means working in a committee-ish environment, with principal developers not on the payroll and thus not able to be whipped at a certain pace, and without a dictator's say in the direction of the product.

That's certainly unbiased wording.

SCdF3• 13 years ago

No. They want to have control, and they think they are large enough to get away with it.

They're probably right.

jfoster• 13 years ago

Canonical's explanation is on this page: https://wiki.ubuntu.com/Mir/Spec

comex2• 13 years ago

It's a short, vague explanation that, other than a correction for a significant inaccuracy in the original version (which I think demonstrates a serious lack of research!), has had no update since it was originally published several months ago, despite the almost universal criticism of Canonical's reasoning by the community. I don't think this is particularly coherent.

j_s2• 13 years ago

Perhaps the ability to re-license for commercial use?

http://mjg59.dreamwidth.org/25376.html

  > They can relicense Mir (and any other GPLv3 projects they own) under licenses 
  > that keep their hardware partners happy, and they can ship in the phone market

mindstab• 13 years ago

Good, they aren't going to have a release that is a massive regression (again). I was planning on skipping 13.10 because everything so far said it would have worse performance and more bugs. OTOH it's a bit scary that their release engineering is making big dramatic structural changes in X servers 3-4 weeks before launch. It should be in frozen locked down beta now... The real solution was that XMir never should have been jammed into their release so forceful while obviously so unready in the first place... Still this is better than launching with it anyways, so half marks Canonical. But this is nothing like the rock solid stability you provided in Beta and Alpha last decade.

Steltek• 13 years ago

Why not? It worked for PulseAudio, right? Or how about the upgrade between the last two LTS releases?

I can't understand Canonical's downright eagerness to force broken software out the door just to appease their roadmap and rigid release schedule. It's pushed as the newbie friendly distro but behaves more like Gentoo when it's time to upgrade.

I cringe when I see people installing Ubuntu on to production servers. The track record just does not support that choice.

jcastro3• 13 years ago

I cringe when I see people installing Ubuntu on to production servers. The track record just does not support that choice.

Wow seriously? Wikipedia, Dropbox, Hulu, NTT, ATT, HP Cloud (Guest AND Host), Comcast. Netflix is moving it's entire infrastructure to Ubuntu. Hate on linux sound all you want but Ubuntu Server is awesome.

Jach• 13 years ago

I'm not sure why you're throwing Gentoo under the bus when its upgrade system is completely different. There is no "time to upgrade" because there are no releases to upgrade to -- you just install new versions of the software you happen to use when they come out and you feel like it. My own experiences with keeping my Gentoo systems up-to-date have been considerably more enjoyable than keeping my Ubuntu systems up-to-date. (I can no longer keep my GUI-reliant Ubuntu systems up-to-date because they're still using 10.04 as I refuse to "upgrade" past Gnome2. So it was bye-bye Ubuntu for me for new installs.)

I agree with jcastro that Ubuntu Server is good, it's easy. It's a popular choice for Amazon ec2 images. It also represents a different mindset in system administration compared to the classic one that went like "Install CentOS version X and keep it for 5 years, sure the software is old as crap even when version X just came out and we hope we can apply security patches without needing a whole new version that depends on something the package manager can't provide, but it's stable!!!" (Just for fun I googled "centos packages old as crap" and this popped up as the first link. http://scalability.org/?p=6270 Notice how it's from two months ago? It's not a new phenomenon either.)

keithpeter5• 13 years ago

"I cringe when I see people installing Ubuntu on to production servers. The track record just does not support that choice."

I would appreciate a pointer to the track record. Genuinely interested not provoking here. Ubuntu Server LTS has always had a 5 year support period, so 10.04 supported until April 2015 and 12.04 until April 2017. Seems widely used.

SkyMarshal• 13 years ago

I have zero problem with Canonical taking risks with their non-LTS desktop releases. That's what they've always been for.

If you can't abide the issues you'll run into and don't want to contribute to the bug hunting, stick with the LTS desktop or LTS server, that's what they're for.

This is a free OS, hence "customers" are also beta testers. Getting new features into their customers/testers hands early and in imperfect condition is the best way to push things forward to the next stable LTS. Learn it, live it, love it.

keithpeter5• 13 years ago

"OTOH it's a bit scary that their release engineering is making big dramatic structural changes in X servers 3-4 weeks before launch"

My understanding is that both Mir/xMir and X11 are in there, just X11 is the default now instead of Mir/xMir because latter not ready for prime time.

Have I got that wrong?

codex• 13 years ago

Canonical is desperate to differentiate and this make money. However, when all the code is open and free, there is no way to differentiate for the better. It's like selling insurance. Everyone's product is the same.

jared314• 13 years ago

A new display server does not differentiate an OS in anyway that would remotely make money. It is more likely they tried to use a technical solution to solve a non-technical problem (display server consensus on an old architecture).

rodgerd2• 13 years ago

A new display server under GPL3 with mandatory CLA would be an excellent way of making money if you got enough leverage that you could start charging hardware vendors if they wanted a private copy so they could avoid disclosing their driver modifications.

It's same relicensing model MySQL made money with.

regularfry• 13 years ago

Assuming they make money on phones, it is. They can say "Look, mobile app development is just desktop app development! No special environments! Just cross-compile the Exact Same Code(tm)!" or maybe "It's Just Javascript!(tm)" and all of a sudden they've got a ready-made ecosystem of app developers who don't even need to install any new tools other than what they've already got installed on their desktops to provide content for their phones. More content -> more sales -> more money, and they must be absolutely desperate for anything that might draw developers' attention away from iOS and Android. If you start from the viewpoint that you need to provide a phone-specific display server anyway, you can see the thinking.

There are bits I think are wrong with this plan, but it's coherent.

shmerl3• 13 years ago

I still hope at some point Canonical will come to their senses and will dump Mir for Wayland. Otherwise this mess will only increase in the future.

A related question. What will SteamOS use after X.org? Wayland or Mir?

pachydermic2• 13 years ago

Excellent question.

But if they really are going to be a healthy contributor to the whole community, then it makes sense for them to pick one and contribute to it. Who knows what's actually going to happen! I don't want to get my hopes up, though. Graphics support sucks in Linux for (presumably) some economic reason - maybe Valve entering the market is just a hedge against Windows locking down their software market and they don't really care about Linux or at least not about the wider community.

But Gaben has said some pretty promising things. Who knows. Very interesting and exciting, isn't it?

mackal4• 13 years ago

A little birdy informed me that SteamOS is based on Ubuntu, they didn't go into more details besides saying it was a dumb choice. I would assume its based on X.org since neither Wayland nor Mir are mature enough.

epsylon2• 13 years ago

it was a dumb choice On the contrary, I think it's a great choice. Ubuntu has one of the best, if not the best out-of-the-box experience, especially for unexperienced users. It's not a coincidence that it's been adopted by many big cos (Google comes to mind) or administrations. In particular, hardware support will be key to the success of SteamOS and there are only few contenders that compete with the friendliness of Ubuntu (especially a Valve-customized one).

krakensden• 13 years ago

Everyone thinks replacing X.org is a clear and obvious win until they try and do it.

The only reason to think differently about Wayland is that a plurality of X.org developers seem to be involved.

Mikeb852• 13 years ago

They should have stuck with Wayland. Gnome 3.10 supports Wayland, and Fedora 20 will ship with Wayland as an option. openSUSE 13.1 will also ship with Gnome 3.10, so getting Wayland working there shouldn't be hard, and I'm sure the Arch and Gentoo communities will have it working very soon too.

Jasper_2• 13 years ago

As a GNOME developer, there's still a lot of work to do. It's mentioned as a tech preview for a reason. We'll work on shipping with full Wayland support by 3.12, but don't expect it to be fully usable or represent the final state of Wayland support.

dylan-m• 13 years ago

GNOME 3.10 doesn't completely support Wayland, yet. Parts of the software stack do, so you can run Weston in a window and then run individual GTK+3 apps with the Wayland backend and they'll probably work, but actual end-to-end Wayland is aimed at a later date. It is getting close, though :)

SCdF3• 13 years ago

I really want Mir to die. Unless I'm misunderstanding the situation Wayland vs. Mir is going to be a massive headache for linux, and afaict Mir exists for no technical reason; it exists because Canonical don't want to play in anyone else's sandpit.

spongle3• 13 years ago

Well by killing XMir they've done that. There is a lot of legacy xlib stuff floating around that will now be excluded from the Ubuntu platform.

Mir is just a square wheel like all the other crap I have to uninstall when someone installs Ubuntu.

I'm back on debian myself now.

vidarh10• 13 years ago

They've not killed XMir, and they've not excluded anything. They've delayed shipping XMir and Mir for the Ubuntu desktop releases until they can sort out some remaining issues.

makomk• 13 years ago

Last I heard XWayland didn't actually work properly either, and the official position of the Wayland devs was that it was Not Their Problem because it wasn't part of Wayland, they had no interest in fixing it, and anyone who suggested this reflected badly on Wayland was just spreading FUD because it wasn't part of Wayland.

voyou• 13 years ago

You've misunderstood this decision. They've delayed XMir and Mir. Ubuntu 13.10 will ship with X as its display server, just like all previous versions of Ubuntu.

devx3• 13 years ago

I don't know much about Wayland, but I know Mir is trying to be very efficient performance and battery life-wise even for smartphones, since it will be based on OpenGL ES 2.0. Can Wayland do the same?

shmerl3• 13 years ago

Wayland already does the same. Sailfish OS which is going in production in the end of 2013 in Jolla's handset is using Wayland.

https://sailfishos.org/

vidarh10• 13 years ago

I doubt it is going to matter much. The major toolkits will continue to support X11 backends for the foreseeable future, and systems with both Wayland and Mir will be shipping with X servers by default for the foreseeable future and likely with X servers as optional packages "forever" afterwards.

It is going to be an issue for people doing low-level/backend GUI stuff and pretty much nobody else.

antocv2• 13 years ago

It is going to be an issue for everybody else because when things dont work as they should, and they wont because now there is two backends to support for the toolkits, it will be Wayland here and Mir there, completely useless for the user, but what will you say in the error message "Something stopped working, sorry?".

Just like alsa, oss, jack, esd and pulseaudio.

I never wanted to touch pulseaudio with a ten-foot pole, it is kept away from my systems in every form. Yet Ive seen apps complain about pulseaudio and have it in their settings by default, then not working and telling me to fix my pulseaudio installation.

There will be apps developed by Ubuntu for Mir that will just cause problems when others try to run them on Wayland.

auvrw• 13 years ago

It is going to be an issue for people doing low-level/backend GUI stuff and pretty much nobody else.

or people who write so-called "suckless" software; most of the things i run on my desktop (notable exceptions including my browser, anki, and perhaps one or two other programs) directly include xlib / import xlib bindings rather than gtk, qt, etc.

jcastro3• 13 years ago

Wayland vs. Mir is going to be a massive headache for linux

Not really. You know what other Linux doesn't ship Wayland, Mir, or X?

Android. And the world hasn't exploded.

misframer• 13 years ago

Apples and oranges.

Touche• 13 years ago

Yet no one else uses SurfaceFlinger because Google's no-patches and code-dump strategies makes it completely useless to any potential downstream project.

Pxtl2• 13 years ago

Notice how few Linux OSS applications have been ported to Android. Where's LibreOffice for android? Inkscape? Blender? Audacity? GIMP? Thunderbird? Filezilla?

Obviously in most of those cases a naive port of the GUI would only be useful for a large tablet with a keyboard and mouse attached. But still, the naive port would be better than nothing.

The near-total lack of direct ports between Linux and Android (except for SDL-based games) should tell you that this does cause a problem.

SCdF3• 13 years ago

My apologies, I guess I needed to be more specific here: I meant desktop linux users. People who use Ubuntu, Arch, Debian etc.

Clearly it's not going to matter for servers either, or raspberry pi, or routers, or... lots of things.

sbuk• 13 years ago

I thought that one of the oft-cited benefits of OSS is choice; surely the existence of both Wayland and Mir offers this?

Edit: Re-reading this, it comes of as quite sarcastic, almost patronising. That is not the intent.

quink• 13 years ago

But I'm confused now!

http://www.theregister.co.uk/2013/07/11/ubuntu_1310_to_ship_... said the following three months ago:

Shuttleworth said he has been running Mir on his own laptop, an "all-Intel" Dell XPS, for two weeks, and that barring a few minor glitches, the system feels smoother than it did before.

Surely it must have been ready to ship in a major release of the most popular Linux distribution given that!

There were only minor glitches and it's run a whole two weeks on the damn founder's personal laptop! How could it have been declared not ready to ship three months, such a vast period of time, later!

Please grow up, Canonical. Also, Wayland.

YokoZar2• 13 years ago

The main problem area blocking the release is multi-monitor support. A regression there means it will break for some users, even though those who don't use this feature might have a huge improvement (such as mark did).

devx3• 13 years ago

I wonder if this has anything to do with Intel's decision to stop supporting Mir. Might've set them back just enough so they won't be ready to ship it on time for 13.10.

mhw• 13 years ago

Your question is answered in the Q&A page linked from the email:

Has the recent revert from Intel triggered this decision?

A: No. It is completely unrelated. This decision was made by our engineering management team and was based solely on the current quality feature set we defined for XMir, independent of any discussions with Intel.

jbeard4• 13 years ago

Does anyone know which graphics cards Mir is intended to support on release?

methodin• 13 years ago

People say competition is a good thing, which I agree with. Competition is arguably the lynchpin of capitalism that keeps us from a slew of monopolies in each market.

It appears however that personal vendettas trump that sentiment based on the posts here.

XMir postponed in Ubuntu

Related Stories