The Story of Ubuntu's Mir Abstraction Layer (MirAL)

[TheBlackCat]

Wayland does not guarantee a stable API or ABI for client developers.

Again, this is simply false. Wayland most certain does provide a stable API and ABI for both client developers (through libwayland-client) and for compositor developers (through libwayland-server).

[darkblu]

Wayland enabled it before that, for Sailfish. So it's not like Ubuntu touch didn't have a working use case to learn from. They literally had no need to reinvent the wheel.

Incorrect. Ubuntu Touch used (without even crediting first), all the work that went into libhybris, which enabled Wayland based systems to run on Android HAL before Ubuntu Touch. That was really not nice of Canonical.

<snip>
* https://mer-project.blogspot.com/201...u-drivers.html
* https://mer-project.blogspot.com/201...u-drivers.html

Ok, that was a good read and news to me - thanks for the links. I, too, as those 'posters around the web' had no idea of the origins of libhybris, and that it was essentially made for enabling Wayland over Android EGL stack. The irony on me is that I've used both said Android EGL stack natively, as well as used Wayland (natively) and Mir (over libhybris), but had no idea of the genealogy of it all : / I agree Canonical could have been much more open about it.

[hsivonen]

Qt, GTK+, libSDL1.2, libSDL2, GLFW, and a number of other toolkits have been ported to libmirclient, just like they have libwayland-client.

If all those toolkits can abstract over both Wayland and Mir, what does Mir bring to the table? How does e.g. a Gtk+ app become better when talks to the compositor using the Mir protocol instead of the Wayland protocol?

I get it that different desktops have their own compositors in the Wayland world. Seems like Unity could have its own. It seems that Canonical values having a stability point at client library ABI. Seems sensible.

I really don't get what having a non-Wayland protocol buys that's so good that it justifies the damage to the desktop Linux ecosystem from the bifurcation. Why doesn't Canonical use the Wayland protocol between a Unity-specific compositor and a Canonical-blessed ABI-stable client library?

[Scias]

They are already doing what they want. It doesn't mean they can't be criticized, when what they are doing is actually damaging Linux desktop as a whole.

Don't get me wrong. I also think Mir is one of the biggest wastes of energy and code in the Linux world but arguing on Phoronix forums about it won't change anything. I'm still hopeful that in 2-3 years they finally drop it and make Unity 8 a Wayland compositor once they realise that still nobody outside them cares about it and that's it's a big extra maintenance burden.

Anyways, only time will tell.

[jo-erlend]

It seems slightly ironic that the people who are most worried about wasting time and resources, are the ones who are spending all this time opposing Free Software they're not interested in, rather than promoting Free Software they are interested in. Wouldn't that time be better spent writing articles about how Wayland works and stuff like that? Seriously; look at the people commenting on Mir articles against Mir and look for them adding something to an article about Wayland. It seems to me that the people who do write interesting articles or comments about Wayland, are not the same people who are constantly attacking Mir, because they're focusing on Wayland. On the other hand, the people who are writing interesting articles and comments about Mir, doesn't attack Wayland, because they're focusing on Mir. That leaves me to suspect that the most hardened Enemies of Mir are usually not all that qualified to participate in a constructive debate about this. Personally, I'm quite enthusiastic about Snap, Unity and Mir, so I spend my time learning about those technologies. I don't spend my time attacking Flatpak, Gnome Shell and Wayland. I like all of those as well, I just like the others more. The Free Software community is never going to evolve into a Single Hierarchy, which to my mind is a good thing.

[Vistaus]

On the other hand, the people who are writing interesting articles and comments about Mir, doesn't attack Wayland, because they're focusing on Mir.

Except for bregma who focuses on Mir and seems to know a lot of positive things about Mir but makes false statements about Wayland, thus having to be corrected by a lot of us.

The Free Software community is never going to evolve into a Single Hierarchy, which to my mind is a good thing.

Agreed. But that wouldn't be the case either with Wayland. If it was, then the Elementary, GNOME, etc. teams would've teamed up already because they'd all be using the same window manager. In reality, they're all writing their own one, proving that even with the same protocol the result doesn't have to be a Single Hierarchy.

It seems slightly ironic that the people who are most worried about wasting time and resources, are the ones who are spending all this time opposing Free Software they're not interested in, rather than promoting Free Software they are interested in.

Just because you see us debating here doesn't mean that's all we ever do, you know. I might not be promoting Wayland as much but I *am* promoting other Free Software outside of Phoronix (like on Twitter, for example). And I'm sure I'm not the only one.

(edit: and in my case, you know that I do spend time outside of Phoronix 'cause you and I, Jo, have debated on OMG! Ubuntu multiple times ;-))

[bkor]

It seems slightly ironic that the people who are most worried about wasting time and resources

Mir is causing a lot of additional work. That's been said for various years. The benefit of Mir: still hasn't been answered. Arguing about it and spending lots of time: that's what I decide to do or not to do. For Mir, things get complicated while the "why?" is not answered. The complication is unavoidable, cannot decide just not to do anything. Then there's the bullshit, like saying 'yeah toolkit will handle it'. I've tried native Firefox on Wayland. Lot of obvious things which are not working properly. Then eventually it all has to be checked again on Mir. Waste.

Now libweston is being made to ease the creation for making Wayland compostors/window managers. As a result, why even have Mir? They (Wayland devs) initially didn't realize that something like libweston would be helpful. Despite that late realization, libweston probably will get ready enough end of 2017. Mir/Unity 8 might finally be used end of 2017.

If there are issues with Mir I expect Canonical half truths such as: "ah but the app developers aren't using the toolkit properly", "ah but the toolkit didn't abstract things properly". One response here already is "yeah but they still make direct X calls".

[Khrundel]

Mir is causing a lot of additional work. That's been said for various years. The benefit of Mir: still hasn't been answered.
...
One response here already is "yeah but they still make direct X calls".

All of these applies to the Wayland too. It causing additional work, and toolkits can't hide all differences (you mentioned firefox), it doesn't bring any visible advantages, it will harm users, there will be tons of problems with outdated apps, drivers etc. Why not stick with X11?

I think, if there will be a painful transition from X to something, it better be that something not made of shit and sticks. And the only way to ensure quality is by comparison. So, if Canonical wishes to spend time and money working on alternative solution, it is not wise to discourage them.

[jo-erlend]

Mir is causing a lot of additional work. That's been said for various years. The benefit of Mir: still hasn't been answered. Arguing about it and spending lots of time: that's what I decide to do or not to do.

So you have the freedom to choose how you spend your own time, huh? Has the benefit of that been answered?

[TheBlackCat]

So you have the freedom to choose how you spend your own time, huh? Has the benefit of that been answered?

Mir devs have the freedom to work on what we want, but we have the freedom to disagree verbally with their decision. No one is saying that they can't work on what they want, but we are allowed to talk about whether working on it actually benefits anyone.