Discussion on:

Canonical axing X Windows: What will it mean for the next version of Ubuntu?

Remember that everything King Midas touched turned to gold, so he was unable to eat any food and starved to death.

Shuttleworth may make Linux more marketable, but it may end up being nothing more than an alter Windows. Gold in a demo, gold in his pockets, but unusable by geeks.

was cryptic
If I understand correctly, Ubuntu is beginning to look like a fork from Linux itself, with very little to do with the rest of the Linux tree.
If it retains its opensource commitments, will it really matter if it's no longer Linux-proper? It would be an opensource competitor to the mainstream proprietaries (MsWin and OSX), so wouldn't that still be a gain? Even if geeks can't use it

That was exactly what ran through my mind, and I am now reviewing my attitude to Ubuntu. I deliberately ordered a bare metal machine for portable computing and installed Ubuntu on it. The announcements WRT GNOME and XWindows are likely to make me discontinue Ubuntu from now on. I've only been trying different flavours of Linux for about 4 years, on and off, but am certain Shuttleworth's decisions are likely to turn Ubuntu into an evolutionary cul-de-sac.

Don't like Unity? Want GNOME and X back? No problem! It's STILL Linux. Download the source, build and install whatever you want. Doesn't come with those? So what? For the longest time, all you got was X and you had to download source to build whatever window manager you wanted.

Nothing has actually changed.

Good grief. I never thought I'd hear geeks whining about Linux not coming with something right out of the box.

You'd hardly call me a geek if you knew me. Sure, I have been using computers for a very, very long time, and depend on them for a great deal. However, the Linux-Ubuntu problem is for me one of remaining within a centrally developed community with a common pathway. If these intended developments had resulted from a long debate, and if the conclusions had a degree of inevitability and agreement attached to them, I might feel differently about the affair. However, and as someone has pointed out, it smells almost like a new proprietary route. One point about mainstream Linux is that it is unlikely to be anything other than a public project, not an owned one. Another is that, although there are variations between different flavours of Linux, they all have core features in common, and it is not so very difficult for users to hop from one to another. My fear is that Ubuntu, in seemingly adopting a unique pathway, may be diverging from the main public pathway. What the motives are I don't know, but already you can see people in this thread sniffing the air.

The reason Linux has struggled for so long is that it mainly appealed to people ("geeks") who mainly wanted to mess around under the hood. Or to people who wanted to get something for free without adequate regard to how it affected work flow. If Linux can become a product for people who mainly concern themselves with ease of use and ability to maintain work flow it finally has a chance to become a significant factor in the market for those of us who have to get work done rather than do a hobby. Or get a headache for free (rather than pay $300 for one as in MS)

Linux is really useful for all kinds of things. The flexibility and the generous licensing are the biggest reasons. It's quite amazing how many network appliances and gadgets are running Linux of BSD. I've only seen two commercial devices running on some non-retail variant of Windows.

Linux is particularly welcoming to embedded systems designers, appliance manufacturers, scientific researchers, and as well anyone interested in running a server, developing software, or just inclined to tinker.

While you can argue these are all "geeky" uses, it may be worth reconsidering your classification. I'm thinking more along the lines of "geared toward academia, research, science, and minimal environments."

I was drawn to it because I like to explore. "Gee, an OS that isn't Windows. I'd like to see that." Same reason I tried BeOS (and still think it's the best OS -- EVAR!), and I'm sure many here were attracted for the same reason.

I use it now because the shell is extremely efficient, and as an IT person, the whole environment is extremely accommodating to the kinds of tasks I need to do, daily.

But, it's no secret that General Purpose Desktop is waaaaay down the list on ideal uses for Linux. Despite the fact that I love it, I still put up with a lot of instability. (But I use KDE 4, so.. really only myself to blame.)

Interesting, but not necessarily the truth. In France the Gendarmerie Nationale has switched from MS to Linux:

http://apcmag.com/french-police-switch-from-windows-to-linux.htm

http://www.canonical.com/about-canonical/resources/case-studies/french-national-police-force-saves-%E2%82%AC2-million-year-ubuntu

The Dutch police will switch to Linux wherever possible:

http://www.osor.eu/news/nl-police-forces-to-use-open-source-software-where-possible

Some of our Police forces in the UK are migrating to Linux, but my brief search hasn't found the example that I wanted:

http://www.theregister.co.uk/2002/10/18/secure_linux_desktop_begins_shipping/

I know of other organisations, but I don't think that I need to produce any more proof than I have, other than to say my latest mobile computing device arrived with no OS. I installed Ubuntu onto it, having trialled it on the predecessor lump before it died. The predecessor worked out of the box, no problem, right down to the wireless networking driver. I have to hack the driver for the new one, but I otherwise amazed a colleague last night when we went through a critical spreadsheet and I told him this was not Excel but Open Orifice.

My choice of equipment has been a netbook, simply because lugging a large device around can be tiresome on the road, and I have a lot else besides to carry. It seems to work as efficiently and quickly as its notebook predecessor. I have installed a package that allows R/W to NTFS portable drives and, thus far, my relatively inexpensive experiment with Linux has been interesting and untroubled. The recent ripple on the Canonical front has me looking for alternatives, and I'll make a bootable USB for (e.g.) Knoppix, and try a few others.

To return to the point, I am not interested in an exercise in messing 'under the hood', nor are many other serious adopters. Nor am I merely interested in a cheap alternative to MS. I am interested in something that is secure, efficient, cheap, easy to dump so that I can switch to another Linux platform, and a few other things are beginning to emerge, such as Ubuntu One. Easy access to compatible packages such as Open Office is a bonus, though I am likely to continue using Lotus because I have come to like the suite, having used it for almost 20 years. I am of course interested in the Lotus/IBM 'Symphony' suite that runs on Linux.

So, in reply to your 'messing about in boats' riposte, no; I have no time for games, and neither do the police forces for which I have cited data. This may be your anecdotal opinion, but anecdotes are like recta; everyone has one. Data, collected using a sensible design or search, are a better basis for generalisation than anecdotes, and I say this as a scientist.

HTH.

...so I appreciate your post, and a little dig all computing all around. Thanks!

I don't think you'll have to build X or the desktop you want from source, as it probably still we be available through the repositories. Probably you'll even be able to choose what you want when you install Ubuntu.

I have to think that if this thing is not going to be compatible with existing apps than it is going to be a major problem to everyone. Who wants to chuck all of their favorite apps because some one decides to reinvent the wheel?

If they can make it so that it will emulate X-Windows and be able run apps written for X than that is a different story.

"The clients can be traditional applications, X servers (rootless or fullscreen) or other display servers."

They will likely make a very very easy to use "legacy" X for "backwards compatibility." Wayland simply changes the stack's organization, rather than removing parts of the stack. It moves the compositor to the bottom over the terminal, then the X server, or other display servers/applications. Apps written to take advantage of Wayland's properties will have some tremendous advantages over the old X apps, but with a nested X, the old stuff should work without modification.

"Who wants to chuck all of their favorite apps because some one decides to reinvent the wheel?"

That's a question Windows users have been asking for years.

rather than have to find 57 different new apps, i would prolly go for 1 new os

and this concern is that compositors historically do not work well with ED games and video players. As far as I remember, I always had to turn the compositor off for the best video playback.

Otherwise, I guess this is a necessary step.

I wish Ubuntu to be like Windows - THE platform. I guess it will be much easier to maintain an other distribution package or to install software myself if there is a working
package to start with, not the mess we have now.

There must be THE Linux distribution, and currently Ubuntu is the best candidate for that.

I have loaded several friends computers with Ubuntu. All I show them is Synaptic Package Manager and how to install and remove a program. My Grandmother and most of my friends do not even know what "Installing the Source package" means and they do not want to learn. Question: Are we saying good-by to "It just Works"?

Given all the foregoing, wouldn't it make sense to invest the effort in generating a leaner, meaner, more secure X Windows than obsoleting a ton of good open-source software? And as for Shuttleworth's Midas touch, I get a leaden feel when I think of Unity (AKA, Pee-Wee's playground) and of a forked-up Ubuntu with everything strained through Wayland.
It's also beginning to look like it's no coincidence that Shuttleworth's initials are the same as Microsoft's.

I excerpt from the 1st page:

"Wayland is a protocol for a compositor to talk to its clients as well as a C library implementation of that protocol. The compositor can be a standalone display server running on Linux kernel modesetting and evdev input devices, an X application, or a wayland client itself. The clients can be traditional applications, X servers (rootless or fullscreen) or other display servers.

The wayland repository includes a compositor which can run as an X client or under Linux KMS and a few clients, but both the compositor and clients are essentially test cases."

The architecture page shows it working with Xwindows with the same number of calls as the normal Xwindows app.

Linux is modular. You can exchange any piece for any other piece. As long as the protocols are standardized, who cares if you're running XOrg, XFree86, Wayland, or whatever?

Now, the fun part is when the code is SUPPOSED to be portable, but then relies on something proprietary. No one realizes because everyone is using the de facto implementation. I suspect we'll see that happen.

I, for one, say "Take my X. Please!" I know all you Buntoos out there probably never have to worry about such things, but in Gentoo land, where we get to do things by hand, getting X running is not trivial. It's "configurationless" these days. Sure it is. That's part of why I have to script my Xrandr settings to run after KDE starts. Or how I have to (or would, if I weren't too lazy to bother) write scripts to detect external monitor hotplug events and react accordingly.

Don't get me wrong, I love that these apps are extensible enough to let me tinker with their behavior in these scenarios, but it would be great to just Do The Right Thing on any display setup, on any graphics vendor's hardware, without looking up guides on whether I should be using Xinerama or not, which GL implementation to go with, or if I should just stick to software rendering for some sleazy Intel chip with brain-dead 3D support.

In that regard, X and its drivers have failed. Miserably.

Are they the target audience for this distro? Aren't there plenty of other distros aimed at them? Must all Linux distros be acceptable to geeks?

I swear to go XWindows crashes more often than every program on Linux combined. In early days of KDE4, just right clicking on the bar could crash xwindows.

But every application relies on it.

I also wonder how this will affect the ubuntu fork Mint, They may have to fork off again if it gets too strange.

Suddenly now we have to move back to Mandriva and the weaker RPM system.

XWindows is pathetically bad. The *only* reason to keep using it is backwards compatibility. Of course, that's a REALLY good reason. "Back in the day" when I was doing desktop support, our XTerms crashed MUCH more often than the Win95 machines. Later on, my first post-college job had me using XWindows on Solaris... and it was significantly less stable than Windows 2000, even though all I was doing was text editors and Perl interpreters. Even later on, all of my negative experiences with desktop *Nix revolved around XWindows stuff... like the machine that refused to remember its desktop resolution settings.

Again, if backwards compatibility is meaningless, than by all means, ditch XWindows. But then again, it's compatibility which keeps me on Windows 7 instead of being able to seriously experiment with other OS's for desktop usage...

J.Ja

You can run X11 as a service under/beside Wayland. If you need X11 compatibility then do like osX and drop it in along side.

(I'm a ringer here after watching the topic shredded in discussion yesterday)

Maybe this will lead some Linux advocates to understand why backwards compatibility is such a stumbling block for each new version of Windows.

We are talking a different approach to the desktop... leaning toward what is happening in world of mobile computing (I am writing this on my Nokia N900 running Maemo 5 - on top linux). Like Maemo, Meego and others. it's just another approach....and probably a good idea.

I like what Shuttleworth has done with Ubuntu, although I am a Mint user. Mint is fork of Ubuntu but they also have a Debian version (not using ubuntu), as well as LXDE, Fluxbox & KDE versions (all built on X).

As long as it remains 'open', I think it could be a good thing....

If you don't like.... you can be a "true geek" and go back to using the cli only!

Remember Linux is the Kernel, Shuttleworth isn't forking the kernel. He's just changing some of the software he packages with the Kernel. Albeit it is radically different than other distributions, but it's definitely not a fork of the Linux.

Let's face it. One of the biggest inhibitors for Linux adoption is game support, and that primarily because of the lack of a good graphics interface. XWindows is fine for basic applications, but it doesn't have the support for the high-end graphics that modern games demands. If they can make the new windowing system attractive to game developers, it will breathe a whole new life into Linux.

Just like DirectX is a layer over windows. Games aren't manually drawing triangles and blitting them to the screen. They just fill a frame buffer and call a flip command. DX handles the rest.

You still have to get the game developers to drop DirectX. The beauty is that they build for DX and the one game blob plugs into any OS with DX.

We have a good graphics interface. OpenGL is cross platform and was doing direct graphics before DirectX came along. What you need is the rest of the DirectX like cradle. If your going to pry away the DirectX game development houses, you need a framework they can mindlessly drop there game blob into.

I think hardware support is part of the problem; advanced features in Nvidia only available to Windows Nvidia drivers and similar situations. That's the ages old hardware vendor versus OS choice issue though. For games specifically, your looking at the framework to make development easy and the politics of convincing game developers that *nix users have wallets also.

"For games specifically, your looking at ... the politics of convincing game developers that *nix users have wallets also."

And that there are enough of them to be worth the development costs.

full screen boogie and old m$ XP discs. no game issues.

Curious as to what makes rpm weaker than any other manager?

Way back when I was looking into which direction I was going to invest the effort to learn I settled on rpm.

Then it was a matter or red hat or Mandrake, and the latter won hands down.

But wherever I landed it was entirely a matter of which package management it was built on, and rpm looked best to me.

I can't imagine not having urpmi an F12 away.

Read, detect, and automatically download and repair all installed packages.

Deb has one, I haven't found one for RPM yet.

Sinister has a very nicely hosed Mandriva install. No bad sectors on the drive but something clearly broken in X or the window manager.


With .DEB based distributions:
- debsum to identify changed packages then aptitude reinstall packagename
- or, edit and rerun build script with "aptitude install" changed to "aptitude reinstall"
- or, dump a list of installed package names and add "aptitude reinstall" infront of them


With .RPM based Mandriva:
- rpm can verify installed packages but the output does not lend itself to recognizing and reinstalling broken packages
- rpm can be used to dump a list of installed packages but names to not match actual package names. You can't rpm > packagelist.sh, add "rpm --reinstall" in front and have it just reinstall without a load of "package name not found" errors. Even when processing the dumped names based on "this is how you get package names from rpm dump" with piping through sed.

Xwindows config has been removed leaving X to auto-config. Xwindows config has been replaced. VM is running under Virtualbox so graphics driver issues are not due to funky hardware.

The issue affects all user accounts as my freshly added user with nothing but /etc/skel/* in it's home directory resulted in the same broken X display.

Anything identifiably X was reinstalled though this is probably very little of what Mandriva drags in for X. As much or KDE that was recognizable by file name was reinstalled. No affect.

Given a way to identify broken packages or binaries, I'd normally do a quick reinstall or purge/reinstall and be done. Given no way to identify the broken package, I'd be able to easily identify the X related packages and reinstall those. Failing that, I'd be able to reinstall my full package list refreshing all but user specific changes. Given Mandriva with rpm and urpmi tools available; nadda on any of the above.

My dirtiest fix was to install over top with the same LiveCD version stamping the OS back over itself without formatting to destroy user files. This is far ugglier than simply reinstalling the package list of the applicable subset of the package list.

PGit, if you can get the VM image from Sinister and provide a cleaner solution, I'd love to hear it. After years of Mandriva, I'm still a little stuned at how well it's borked.

Now, RPM may be a fantastic package format. I'm told it's on par if not ahead of the DEB package format. RPM related tools are horrid after working with apt-get/aptitude though. Actually, I should check and see if that Mandriva VM has apt tools modified for rpm.

First off, "changed file" doesn't equate with "broken package." Therefore doing what you suggest with deb (which is doable with rpm, granted needs more than a few pipes) is going to hose everything that has had it's config under /etc touched, ie you will bork Peter to fix Paul.

Second, given that you are not really identifying what is actually broken, but merely anything with changed files under the package, you will basically be reinstalling the whole system anyway, the hard way.

You might actually find the system is still broken afterward. Nothing under /var is going to be touched, and what if the problem was something like a bad php file you put somewhere that the package manager won't be looking for anything changed?

Doesn't sound like this is any special magic with deb. Just a near complete reinstall the long, hard way, which is going to break all your servers or other with a config under /etc anyway. Right?

I consulted the Oracle of Cairo (my linux "ok, wtf now!?" go to guy) and he said this same is very doable with rpm, but what's the point? A clean reinstall would be accomplished long before this kind of waste of time would play out on either system, including having to go through all your configs afterward in either case.

The operative point is not 'can you do this?' rather it seems to me there's already the answer: 'why the hell would you want to?'

So logically all that could be damaged is some file is now missing and not working. On my other VM's between 1 and 30 clusters got damaged (Windows VM's) Mandriva was my only Nix VM at the time, it was unrecoverable.

The insult comes from the fact that all the windows VM's were recoverable with no data lose at all without requiring install discs. Mandriva is the only one that doesn't seem to have a 1 word tool to fix file system problems (equivalent of Chkdsk /R).

You can't get to a non graphical log in? It won't even go to single mode?

I also have to ask; there are no snapshots from a previous, working system?

Read his

It has a chkdsk equivalent utility which is not finding any bad sectors in the partition. There are no longer any bad partition sectors; chkdsk reports that the filesystem is clean.

A question may be why win chkdsk was able to recover data from lost sectors while lin chkdsk didn't. For this, we'd have to confirm that the breakage is sectors lost from specific files in addition to confirming that equally critical win files where damaged and recovered.

Either way, we have to know the actual problem before understanding the solution and cause. Hence, someone with more rpm knowledge than I providing the broken package or at least the most efficient fix.

"You can't get to a non graphical log in? It won't even go to single mode?"

It appears to be a breakage in X or KDE. The system loads cleanly to a login prompt and accepts/validates uname/paswd. My first step was to boot to single user and default the system to console boot; there is no issue with existing or new users until startx kicks off the graphic stack and DE.


"I also have to ask; there are no snapshots from a previous, working system?"

Not relevant. The question is "how do I fix my Mandriva install on ext3 which apears to suffered the exact same file system damage as my Windows installed on ntfs which recovered at next booth with the automated chkdsk and related "unclean shutdown detected" processes."

For me, the question is "what broken and how can it be fixed so I understand and can fix similar issues in future" or "Given how I'd fix a .DEB system with same breakage; what is the equivalent process for a .RPM distro?" In all the years I used Mandriva as a primary distro, I didn't see something break like this (of course, I had the advantage of scripted installs and separate /home partitions making for easy reinstalls)

At least in Win7, it said there was critical system damage and it needed the install disc to repair it. This ran WHILE inside Win7.
XP refused to boot, broken system font.


Edit, changed title cause Neon double posted making my post make less sense!

Just log into the CLI and type:

export XSESSION="Xsession"
startx

Use anything in /etc/X11/Sessions for XSESSION.

If X runs TWM (or Gnome or ...) just fine, then you ruled out X. That means KDE's broke.

Welcome to KDE!

Perhaps Neon can provide

First, Ubuntu moving to Unity/Wayland is a year or more away in testing distributions. It's not like the next major version of Ubuntu is going to make this transition and an overnight policy change.

When the base distribution does change, Mint may very well stick with it. Hopefully based on improved hardware support and such. If Ubuntu remains an applicable based for Mint; it will.

If Ubuntu alienates Mint as a result of the distribution changes; Mint will simply start from the Debian distro base resulting in Debian -> Mint instead of Debian -> Ubuntu -> Mint.

That's the guess anyhow.

At least Mint has a beta debian version, perhaps that was a good forsight!

Went to the Wayland website and overviewed the detailed architecture. Looks to be it's own graphic server that X can be plugged into.

Saw this old article that gave added insight: http://www.phoronix.com/scan.php?page=article&item=xorg_wayland&num=1

Also was curious about Fedora and what the noise there was over there. As usual, nothing but a mention. Plymouth attempts to use KMS when the right chipset is detected.

Strangely, Wayland looks to be tied to KMS which is dependent on newer graphics chipsets. So Ubuntu won't be running on older hardware?

I understand where he's going, but this is so monumental, it's bound to fail. Seriously? Abandon X-Windows? If anything, it would be wiser to make it transparently compatible with X-Windows, but make it better in every way with new calls, etc.

What other companies practice this method of business and development? Dare I say Microsoft or Apple.
Where is my extinguisher, let's look at this clearly. Ubuntu browser, Ubuntu this, Ubuntu that, all rewritten to run on a totally different windowing system to the current standard, X. Smells pretty proprietry to me.