Ubuntu 14.04 Codename Revealed, Mir Haters Attacked

[Hamish Wilson]

Ultimately the cream is going to rise to the top.

Tell that to the people who developed Betamax...

Since when Canonical have to worry and take care about other distros projects?

Because Canonical is not its own little island, even if they are trying their best to dig a moat. It should be remembered that without the community Ubuntu would never have existed, and that it has taken a great deal of work from Red Hat, Debian, and others in the past to get to where it is today. If the community had not cared about other distro's projects, then Ubuntu would never have existed.

And despite their spiteful protestations, I still think they can not keep trying to do everything on their own. They simply do not have the resources. So by ignoring the work of others they are harming them as well as themselves. I can not see how any of this can be painted as beneficial.

[V10lator]

Where did the man do the mistake?

Here:

He was looking at their work for some time before he realized, that he needed something slightly different.

Because there is absolutely no technical reason why he needed something different. Also if he really needed it to be slightly different he could have used extensions (Waywater supports extensions).

Do you think Canonical could have integrated Wayland as fast as they're developing Mir?

Yes, even faster. If all the devs working on Mir would have been working on Wayland...

They wouldn't have been able to fix it since they don't control it.

If there's really such a problem: Fork it. If they would have forked Wayland with valid reasons nobody would have a problem.

Correct me if I'm wrong, but is there any effort to port Wayland to Android? If not, then I think Android support is the single most important reason why Mir exists.

Mir uses libhybris to support Android. Libhybris has been developped for Wayland before Mir existed. Stop believing the FUD from Canonical.

At least since Canonical announced Mir, the Wayland team kinda got their asses in gear lol.!

The Wayland devs did not speed up the development since Mir was announced. It (almost) looks like that to uninformed people cause the most work they did the last 5 years was prepaing the graphic stack (kernel, mesa, DRI, ...) for a modern display server like Wayland (and Mir. Canoncial directly depends on that work).

in fact going with TDD/agile on a new codebase has allowed them to have an integrated shell + compositor in a "workable" state after 15 months or so, compared to the 5 years and counting since the inception of the wayland project
besides, as fixated as canonical seems to be with C++ now, they would hardly ever contribute to a (C based) project such as wayland - while mir fits in the picture of progressively building a new (at least as far as code is concerned) and distinct platform one piece at a time

While directly depending on the 5 years of Wayland development. Go give them the graphic stack we had 5 years ago and tell them to base Mir on that, then see how long they'll need...

Why do we have to talk in circles over and over again?

[chrisb]

This is what I don't understand.

Canonical devs: "We're tossing away all the work done by the community and developing our own display server because we want to have more control and to not be tied to a protocol set by others. We're making a display server that is developed for Unity, with Unity's and Ubuntu's needs as a priority. Oh, and we expect that it will become immensely popular and everyone else will adopt it."

Something there doesn't add up. It's like saying, "I developed this engine for my 5-wheeled motorbike that runs on fish oil, I expect it to be hugely popular for use in all gasoline-powered cars."

Exactly. It's like the Wright brothers inventing a three axis control system for their airplane, which then went to on be used by everyone else in every future airplane design.

[chrisb]

Your theory makes more sense if you disregard speed as a reason. They can't sell Wayland to an OEM, but they can sell Mir.

Nothing in the Wayland licence prevents Canonical selling it to OEMs.

[RahulSundaram]

Nothing in the Wayland licence prevents Canonical selling it to OEMs.

Wayland is a protocol. You cannot sell a protocol to anybody and Weston is a test/reference implementation and OEM's don't want a proprietary license for a reference implementation designed to flush out bugs.

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

[mrugiero]

Wayland is a protocol. You cannot sell a protocol to anybody and Weston is a test/reference implementation and OEM's don't want a proprietary license for a reference implementation designed to flush out bugs.

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

Yet, they can do their own Wayland compositor and sell it to OEMs, and IIRC libwayland is MIT, so again, they can sell it to OEMs. Of course, then, they would be more like selling their brand than the actual software, since the OEMs could get the software for free.

[Delgarde]

There is also things Canonical has done PR wise to piss of everyone involved with wayland.
...
B. Mir copied XWayland wholesale and released a big PR about their solution for compatibility on MIR.

Not just Xwayland. Canonical talk about libhybris as if they'd written it as a Mir feature, but that too was originally someone else's work done for Wayland...

[Delgarde]

Mir uses libhybris to support Android. Libhybris has been developped for Wayland before Mir existed. Stop believing the FUD from Canonical.

Yeah, that one's particularly annoying... Canonical keep trying to claim that as a win for Mir (or even as a reason why they chose not to use Wayland), even though the code in question was created by Wayland developers....

[TheBlackCat]

The only time "it isn't a protocol" would matter is if you wanted to create a competing implementation.

Yes, like kwin or mutter. That is the problem, not one can write their own Mir compositor like they can with Wayland. So claims by Mark that kwin would soon work on Mir were outright fabrications.

[TheBlackCat]

Do you think Canonical could have integrated Wayland as fast as they're developing Mir?

Yes. Fedora is shipping a test version of Wayland right now. There is no working desktop implementation of Mir, in fact even their bare-bones X wrapper for Mir isn't even in a usable state yet. If all the work that had gone into Mir had instead gone into Wayland, then yes they almost certainly could have had something at least functional for their upcoming release. They could easily have had something like XMir for XWayland working well enough to enable by default if they had just focused their attention on that.

If they had gone the Wayland route, what if they encountered a snag along the way that proved Wayland unsuitable for their needs? They wouldn't have been able to fix it since they don't control it.

Wayland support extensions exactly for this reason, they could add whatever they need. And they certainly would be able to submit fixes, and worst case scenario maintain downstream patches (like every distro already does for X11). It would be far, far, far faster and less work long-term to go this route than to maintain an entire display server, and toolkit patches, and driver patches downstream, which looks like what they are going to end up having to do.

Correct me if I'm wrong, but is there any effort to port Wayland to Android? If not, then I think Android support is the single most important reason why Mir exists.

This is a myth, a myth that canonical has been very reluctant to try to correct. Canonical did not develop their Android support, they forked an android support library developed for Wayland, then didn't tell anyone about it. Their announcement made it sound like they were the ones who developed it, then once it came out that this was false they haven't gone out of their way to correct their fans' misunderstanding.