Mir

Mir will cause fragmentation in Linux on desktop

Bug #1149581 reported by Simon K on 2013-03-06
76
This bug affects 17 people
Affects Status Importance Assigned to Milestone
Mir
Invalid
Undecided
Unassigned

Bug Description

Since other Linux-Distribution will hold on to wayland, Mir will lead to a heavy fragmentation in Linux.

Why other distributions (unless they are deviates of Ubuntu) won't support Mir:
 - You can't just simple work together with others on Mir, you have to sign a contract to submit mainstream changes.
 - Wayland is far more advanced. (This argument has to become invalid in near future)
 - Mir is centered around Unity(Next), other desktop-environments will get troubles by implementing their solution into Mir (if this is possible at all)

Solution:
Since Mir isn't advanced too far and Wayland has a heavy development ongoing, switch back to Wayland (Build Mir as a replacement for Weston if you like to...)
You could even convince those guys to deliver a "wayland light" (without those problematic parts)

Advantages:
 1. Wayland has many of the "Mir-ToDos" solved already
 2. You can guide the wayland-project to its goal (being *the* replacement of X.org)
 3. Wayland is developed by the *community* [Canonical can step into the development but won't need as much developers as Mir will be taking]
 4. Linux for desktop[which includes Tablets and even Phones] isn't parted as heavily as it will be with the introduction of Mir (see reasons above)
 5. Graphic-Vendors (namely Nvidia, Amd Graphics) will adapt this new standard faster (because it is only *one* standard)

Disadvantages:
 1. You have to contact the people from wayland.
 2. You have to discuss steps which potentially aren't constructive.
 3. Bad press (hey you guys have practice in this^^)
 4. wayland isn't invented by Canonical

PS:
Divide et impera [Divide and rule] - Niccolò Machiavelli (1469–1527)
to address bug #1 Microsoft, Google and Apple will have a much higher chance to continue being market leaders, if (GNU) Linux can't bundle it's forces.

Zisu Andrei (matzipan) wrote :

My belief is that, in order for Unity to excel, it needs to detach from the "legacy of bad interfaces" that is represented by all the other desktop environments and display servers, and start fresh. If you look in the application store right now, you will find that 90% of the applications simply do not match Unity design guidelines or do not even have a friendly design, which is vital if Ubuntu really wants to succeed on this side.

Also, work on Wayland/Weston started in 2008. It is 2013, and, only now has it started to get some momentum. Ubuntu started work on Mir in June 2012 and it is already rolling it out to the public, with the first major release in October 2013.

If you read the https://wiki.ubuntu.com/MirSpec, you will find some of the reasons behind this "rupture", as well as the plans for compatibility with X11/Weston.

I like Ubuntu's boldness and I think it will only make other distros more aware of the need to be competitive.

Changed in mir:
status: New → Opinion
Download full text (3.2 KiB)

> My belief is that, in order for Unity to excel, it needs to detach from
> the "legacy of bad interfaces" that is represented by all the other
> desktop environments and display servers, and start fresh. If you look
> in the application store right now, you will find that 90% of the
> applications simply do not match Unity design guidelines or do not even
> have a friendly design, which is vital if Ubuntu really wants to succeed
> on this side.
The same integration-aspect can be archived by removing every software from Ubuntu Repos [they won't work with Mir nonetheless, so this has to happen in the future]
After this "muck out", Canonical can easily add requirements (Visual and Functional), what the (GUI) applications have to "implement" in order to get into the repos.

> Also, work on Wayland/Weston started in 2008. It is 2013, and, only now
> has it started to get some momentum. Ubuntu started work on Mir in June
> 2012 and it is already rolling it out to the public, with the first
> major release in October 2013.
This is a good point of yours. But if granted: a look into the future:
People developing Mir: 8 [1]
People developing Wayland: 51 (April 2012) [2]

Potential developers behind Mir: Canonical employees ~500 [3], Everyone crazy enough to sign a contract to mainstream OpenSource-Changes.
Potential developers behind Wayland: Readhead employees ~5300 [4], Attachmate Group (SUSE) employees ~750 [5], Every OpenSource developer.

> If you read the https://wiki.ubuntu.com/MirSpec, you will find some of
> the reasons behind this "rupture",
Yes, I read those.

> as well as the plans for compatibility with X11/Weston.
So...
Every application running under X11 or Wayland will be usable by Ubuntu-Users.
*Not* Every app running in Ubuntu will be usable by other Distro users...
Well played Canonical.

This statement in contrast with your first paragraph, Every ugly app will also run in Ubuntu in the future...

> I like Ubuntu's boldness and I think it will only make other distros
> more aware of the need to be competitive.

Yes. I agree with you in this.
However, the main opponent of Ubuntu isn't Fedora nor Suse, it is Microsoft Windows, Google Android and Apple iOS.
Fedora, Suse and Ubuntu are all heading in the same direction in being opponents to the "big 3". We shouldn't fight against each other, we should however try to work together as well as possible.

Mir is an interesting move, I'm concerned -however- that Mir could become to much to be handled by Canonical.
And I think a low-level component (like the display-server, kernel or sound-server) should be consistent on all Linux-systems.
Only if every Linux-Developer is backing those parts, we'll be able to address Bug #1 and archive a "ecosystem" where Main-Vendors/Producers are interested in (Ubuntu) Linux.
Btw. by "backing" I don't mean actively developing on one of those parts. I also include "making software which runs against this infrastructure" [Valve, Nvidia, Amd Graphics to name the big once recently in the news...]

[1] http://www.phoronix.com/scan.php?page=news_item&px=MTMxNzc
[2] http://www.phoronix.com/scan.php?page=news_item&px=MTA4OTM
[3] www.canonical.com/about-canonical
...

Read more...

Michael Shigorin (shigorin) wrote :

> My belief is that, in order for Unity to excel, it needs to detach
> from the "legacy of bad interfaces" that is represented by
> all the other desktop environments and display servers,
> and start fresh.
Aiming too low, ubuntu definitely needs to reinvent the world given the legacy of bad behaviour.

Wish sabdfl would become bold enough to announce what's done not what's being tinkered with -- Canonical and ubuntu fanboys have become the endless source of announcement (and reannouncement, and clarification announcement, and rollback announcement) spam over the years, unfortunately.

inte (inte) wrote :

I always thought the main goal of implementing a replacement for X11 is to eventually get rid of X. However, with all major distributions switching to Wayland except Ubuntu switching to Mir the only compatibility layer between these two will remain... X-windows, since from what I understood apps compiled against Mir will only run in Mir but not in Wayland and vice versa.
Since as well Mir as Wayland both feature an embedded X-server, this fragmentation in Wayland and Mir linux flavours will render the project goals useless, since most (especially the proprietary) apps will certainly still be released for X-windows as that will be the only way to cover both, Linux flavours running Wayland as well as Linux flavours running Mir.
I don`t see much point in supporting proprietary Android drivers either, since these are often linked against certain, sometimes outdated kernel versions, which will lead to a whole bunch of problems regarding updates and security.
Ubuntu should overthink the Mir strategy and at least talk to the Folks from Fedora, SuSE, Tizen and Jolla to somehow agree on EITHER implementing Mir OR Wayland on all distris, but not fragment the Linux world.
Thank you.

no longer affects: ubuntu
Robert Ancell (robert-ancell) wrote :

This is not the right place to discuss this. Bugs are for problems with the Mir code, not for the existence or usage of this project.

Changed in mir:
status: Opinion → Invalid
Daniel van Vugt (vanvugt) wrote :

"Mir will cause fragmentation in Linux on desktop"

It's certainly a valid argument and probably true. However as the debate rages, everyone should keep in mind:

1. The existence of this project is not a "bug" in the code of branch lp:mir so should not have been logged on bugs.launchpad.net.

2. "Linux" is not an operating system. It is a kernel. There are other operating systems based on Linux such as Android.

3. "Wayland" is a protocol and a couple of libraries, but not a compositor/display server. The compositor/display server is "Weston".

4. "Mir" is a compositor/display server. It is also a protocol but could be modified to use the Wayland protocol in theory. It's not impossible but is a whole separate argument about whether it should. That's probably the real argument here.

5. "Canonical" is a commercial enterprise and needs market differentiation like any enterprise. So arguably, some fragmentation is a necessity for Canonical.

7. All of Mir is open source. Can we celebrate that fact a little?

These are my personal opinions and not necessarily the opinions of my employer :)

Daniel van Vugt (vanvugt) wrote :

And apparently I can't count. I know.

alexus (alexus-m) on 2013-03-07
Changed in mir:
status: Invalid → Opinion
status: Opinion → Invalid
inte (inte) wrote :

@Daniel
You are absolutely right, Android is Linux. However, it is not GNU/Linux and not compatible at all.
Ubuntu IS compatible by now. If Canonical planning to somehow fork from GNU/Linux in an Android-like style they are of course free to do so, I would only consider that to be not really helpful for either side.

inte (inte) wrote :

And, regarding market differentiation - is it Ubuntu trying to grep market share from Windows, OSX, Android etc. or is Ubuntu rather trying to become the market leader in the Linux world by differing from GNU/Linux?
If it was the latter I don't agree. The different GNU/Linux flavors should rather join forces to become strong and a reasonable competitor to other OS, not fight each other.

Daniel van Vugt (vanvugt) wrote :

Please see bug 1
:)

inte (inte) wrote :

Ok then we agree at least in your point 4:
Mir NEEDS to be modified to use the Wayland protocol, otherwise the GNU/Linux world will be disputed in Wayland and X, mostly utilized by enterprise systems and in science, and in Ubuntu, mostly used by private customers.
Furthermore, Mir needs to support binary drivers which will eventually appear for Wayland instead of requiring dedicated drivers.

inte (inte) wrote :

So läuft das bei Microsoft:
http://techrights.org/2009/01/22/microsoft-taskforce-vs-walmart-linux/
Die sollten das zehnfache zahlen!

inte (inte) wrote :

Whoops wrong browser tab. Is there any way to delete comments here? Can't figure that out... Thanks!

Simon K (octav14n) wrote :
Download full text (7.0 KiB)

My thoughts on the given comments:
#3 (@shigorin) - isn't constructive (I've asked the author to revisit this post, nothing happened), I'll not waste my time by discussing this.
#4 (@inte)
> I always thought the main goal of implementing a replacement for X11 is to eventually get rid of X. However, with all major distributions switching to Wayland except Ubuntu switching to Mir the only compatibility layer between these two will remain... X-windows, since from what I understood apps compiled against Mir will only run in Mir but not in Wayland and vice versa.
> [...]

This is a serious point. If we have 2 new Systems there are 4 possibilities:
 1. System Weston/Wayland gets support
 2. System Mir gets support
 3. Both systems get support
 4. The minimal criteria/overlapping will get used.

Point 3 is unlikely, no capitalistic company wants to support 2 systems in the long run
Point 4 leads us to a schizophrenic position: The highest level available to both: Mir and Weston/Wayland is: XOrg.
Even more important, if you guys implement a "MirWayland"-Backend, Wayland will be used, it (will be/is) far more advanced than XOrg and I can't see for a WaylandMir-Backend in the long term?!

> I don`t see much point in supporting proprietary Android drivers either, since these are often linked against certain, sometimes outdated kernel versions, which will lead to a whole bunch of problems regarding updates and security.
> [...]

I have a different opinion on this one: Using Android-drivers gives Ubuntu a huge advantage.
Ubuntu was the first platform to properly support Nvidia/Amd proprietary drivers ("Blob"). This has lead (or at least supported) to the high count of Ubuntu-Users. Giving the Blob-Vendors a infrastructure their engagement, quality and support have increased a whole amount.
I appreciate this step. Linux and (GNU)Linux lack of proper interfaces for binary drivers. Every time a new XOrg version or Kernel came out there was a high risk (in case of XOrg updates it even was assured) that old Binary-Drivers didn't work with the running software.
Ubuntu resolves this problem by accepting Android-Drivers. They have (from the Compositer-Viewpoint) a relay good interface.
Using the Google-designed interfaces is good because they exist. They have proven to be usable (Android is the main player on smartphones) and many companies have the drivers ready even before UbuntuPhone is there.

> Ubuntu should overthink the Mir strategy and at least talk to the Folks from Fedora, SuSE, Tizen and Jolla to somehow agree on EITHER implementing Mir OR Wayland on all distris, but not fragment the Linux world.

... approved without a comment needed ...

#5 (@robert-ancell)
> This is not the right place to discuss this. Bugs are for problems with
> the Mir code, not for the existence or usage of this project.

Maybe my headline is misleading (I purposely did this though), But I didn't report a bug against "lp:mir", I reported a bug against "Mir",
Registered 2012-06-20 by Thomas Voß

Mir is a display server technology.

This isn't the code, the whole project has this "Bug", with this discussion succeeding this report will pass through the different stages:
 - Marking...

Read more...

Simon K (octav14n) wrote :

Hm, this is highly misleading:

Am 07.03.2013 21:29, schrieb Simon K:
> This isn't the code, the whole project has this "Bug", with this discussion succeeding this report will pass through the different stages:
> - Marking as Confirmed: It is a issue. (regarding many people)
> - Marking as "Won't fix": The discussion ended with the understanding that Mir is essentially for the future of Ubuntu - Mir will get developed as Canonical planed to (at the time from the start to the publication).
> - Marking as "In progress": The discussion ended with the understanding that we have to use a single Compositor- We will search for a possible solution in this problem.

I meant to say:
This isn't the code, the whole project has this "Bug", with this discussion succeeding this report will pass through the different stages:
 - Marking as Confirmed: It is a issue. (affecting many people)
 - Marking as "Won't fix": The discussion ended with the understanding that Mir(the protocol) is essential for the future of Ubuntu - Mir(the protocol) will get developed as Canonical planed to (at the time from the start to the publication).
 - Marking as "In progress": The discussion ended with the understanding that we have to use a single Compositor-Protocoll- If this is the case, lp:mir has to be touched because of this bugreport.

inte (inte) wrote :

One more comment:
How will all the by-now-compatible Ubuntu forks (such as Kubuntu etc. http://en.wikipedia.org/wiki/List_of_Linux_distributions#Ubuntu-based, or Linux Mint eg.) handle the change of the compositor of the main distribution they rely on? Either they will have to port their DM/DE to Mir as well (which will be a whole bunch of work which is probably impossible especially for the smaller projects) or stay with X-windows. However, keeping with X-windows will result in not being able to run Ubuntu apps (like probably, the Software center). Using Mir compositor and running the whole DE via a nested X-Server in Mir will probably also be an option but that would probably be far less efficient and doesn`t sound very reasonable to me as well. I guess this needs further clarification.
I only wonder why this whole issue isn`t discussed in a brighter scale...

inte (inte) on 2013-03-08
Changed in mir:
status: Invalid → In Progress
Changed in mir:
status: In Progress → Opinion
Fabio Rafael Rosa (rafaelfdr) wrote :

Application wise, it will not make much difference, as apps will target Qt, or Gtk+, or SDL
Toolkits must address the problem, not the apps (at least, not most of them).
Canonical already told the they will provide the support (and handle the burden, because, they create the 'problem', they have to deal with it).
The most sensible point, is drivers.
Up to know, Mir uses KMS/DRM , so, no change at all
My only worry is that wayland/weston and mir share the proprietary vendor drivers (nvidia, ati) .
Up to now, Canonical said that that will be the case (the drivers will be shared). But, it´s just talk, we don´t have _any_ proprietary vendor supporting mir _OR_ wayland.
Despite not liking canonical creating Mir (and how they did create it), they have the right, and I ´ll hold my judgement for when we start to see how things will got with proprietary vendor.
By then, if we see 'Mir only' drivers, then, I´ll agree that Canonical did really turn into a bad Apple (tm) on the basket, uninstall Ubuntu and Bash it until the day I stop breathing, when I´ll tell my son, at my death bed, to keep my legacy (bashing them) :P .
Despite all the crap we saw flying these days, I don´t think that will be the case (at least, I wanna believe that Canonical is better then Apple and Microsoft).

Simon K (octav14n) wrote :

#17 (@fdrs):
Amen^^
True words are spoken here. I would much more prefer it if Mir would be using Wayland...
but as long as we aren't confronted with "Wayland/Mir"-specific software/hardware (which shouldn't be the case because of abstraction-layers as pointed out by you @fdrs ) I'm ok with it.

And the news from Canonical about Mir are giving me the hope that this topic (interoperability) will be addressed at the time of designing/building Mir.

Kind regards
Simon

Market as invalid since it is a flamebait. Please don't reopen it.

Changed in mir:
status: Opinion → Invalid
tags: removed: bug-1 mir wayland weston
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers