Canonical Has Been Weathering The Pandemic Well: Turned A Profit, Back Above 500 Employees

[libv]

I meant "8 years since Mir was started" rather than "8 years since Wayland was", but thanks anyway: I didn't realise it had been that long.

13 years for Wayland though, huh? eesh. And that's only "so far", and as you say, it started from a much better position than X did.

Since you're one of the X devs, and I gather they/you are all working on Wayland now, why HAS it taken such an incredibly long time? Not trying to be snarky: I just don't see how, well, ANYTHING can take anything like that amount of time, let alone a cut-down windowing system. What's the piece that I'm overlooking?

I'm a driver guy (such inconsequential things as modesetting, free ATI driver, ARM GPU drivers). To me X or wayland is just "infrastructure".

From where i sit, Wayland is one of those typical reinvention cases, promising that everything will be golden in 6 months, fixing what was broken on the previous implementation, but throwing everything good away as well. That never works out, as _everything_ gets reinvented, and several features get dropped (remote X anyone?). Aaron Plattners presentation X extension from 2007 was totally ignored for instance, and that should have gotten us some quite steps towards a 3d engine based window rendering model for X.

The only reason that it actually succeeded, for a value thereof, is that Kristian was part of red hat, and then intel, and thus part of the only political class that is allowed to reinvent things. Which is the prime reason why Mir failed, only some people are allowed to do anything major in the world of X/fd.o, everything else will be trampled.

Here i am, 8 years ago (so mir is older than 8ys), explaining how i see the Mir vs Wayland thing, when intel driver developers were forced to remove a tiny bit of Mir driver side support, by intel management: https://libv.livejournal.com/25325.html Intel is a hw company btw, and should care about maximising its driver support across its userbase. They should not limit the applicability of their driver support because of some political games at a much higher software level.

There seems to be a temporal correlation between the above games and Kristian moving away from working on Wayland at Intel, and then later moving to google. I have not been out drinking with Kristian since, so i have had no chance to ask him what happened.

[F.Ultra]

I am not an accounting expert, but they tell me that debts are not synonymous with a company that is doing badly or doing well, there are other parameters to consider. For example, if the customer portfolio increases or decreases.
A company may be in debt, but have made investments that could lead to further gains in profit over the long term.
For example, I know for sure that SUSE recently bought Rancher.

Yes it depends on how and why you got your debt of course, but it also depends on if your debt is decreasing or increasing, at some point the debtors wants their money back. What worries me a bit is that their stock took a dive when they released their first independent report (2021-Q1) so the numbers was not as good as was expected, then on the other hand their share is not highly traded either so not much can be derived from that.

Looking at their reported results though: https://se.tradingview.com/symbols/D...ome-statement/ (click on quarterly to get their Q reports to get more data) one can see that they have reported a negative net income in every report for the last 4 reports and the last one is €-97M, that was what I was referring to in my first post.

[arQon]

From where i sit, Wayland is one of those typical reinvention cases, promising that everything will be golden in 6 months

Many thanks. So, generic CADT basically - and, as you say, only RH and Intel are allowed to do that.

So, 13 years and counting to write a windowing system with only a fraction of X's capabilities, with none of the burden of all the underlying work like drivers etc. That's actually so bad it's almost impressive.

[arQon]

I just noticed that the article you linked (from 8 years ago, and quite possibly where I got the timeframe from) has this little gem in it:

> Long-time X.Org contributor Daniel Stone, who's also been commenting in our forum thread, was quick to say, "The best part is that the input bit of the rationale is totally wrong: there's no way for clients to get another client's input, short of mapping a full-screen transparent window and convincing the compositor to not decorate it.

So, surely, thanks to the stupidity of CSD we can now do exactly that, no?
edit> That's unfair: there are ALWAYS ways to have e.g. borderless windows etc anyway (or at least, there should be). But OTOH, that means Stone's argument that that ISN'T possible with Wayland (or shouldn't be) is either false or indicates a crippling defect in Wayland's design. So either way it's pretty bad: I just don't know which of those two bad outcomes is the correct one. (Or if both now are).

That's rather worrying if correct. Not just for the actual security concerns, but also because if the Wayland fanboys who spam "OMG SERCUTIY MELTDOWN!!11!" in literally every. single. thread. about X were capable of independent thought, the irony here might cause their heads to explode. :P

[AHOY]

I

Maybe it was a hard problem to solve and people made many mistakes in the process? If it was easy, Mir would have just dominated with their implementation. There wasn't anything stopping you from pulling a Microsoft and extinguishing Wayland from the inside with your superior protocol and implementation. I don't like some companies having too much control but the "community" is often all talk and no action. Did Canonical even considered doing something like Wayland/Mir before people started doing Wayland? Why did it fail? Or did they only reacted in an attempt to gain more control? At worst, Wayland paves the way for a better protocol to emerge with all the lessons learned. Especially the non-technical ones and the protocols that show what kind of consensus the different people found tolerable. What flaws they found after the implementation. Wayland is the best way to get rid of Wayland.

Qaron here's a lad you'll fall in love with instantly https://gist.github.com/probonopd/9f...5e3a9f2d1f2277

[AHOY]

I

Can you explain what this was all about? No sarcasm and I'm not trying to imply anything I'm really curious.

https://pastebin.com/KjRm3be1

Found some more context, unfortunately it's reddit

https://old.reddit.com/r/linux/comments/19nehd/canonical_announce_custom_display_server_mir/

https://old.reddit.com/r/linux/comments/19ngad/ubuntu_announces_mir_a_xorgwayland_replacement/

[arQon]

Maybe it was a hard problem to solve and people made many mistakes in the process?

It's not a hard problem, at all. I've written windowing systems from scratch in about as week (albeit simpler ones, certainly) as have plenty of other people. But it's a fully-known problem, and a typical timeframe for such a thing, including a basic toolset, would be MAYBE a MONTH or so for a handful of devs familiar with the concepts - especially since Wayland discards all the pieces of X that are actually "hard". And remember, these are people who were supposed to be experts already because of the work they did on X.

But 13 years is a joke, even for a group that didn't have that experience. Hell, 13 years is a joke even if they had to learn how to *code* first. To take the first comparison that comes to mind: 13 years takes you from Quake1 to real-time photorealistic rendering. And in that time, the Wayland team achieved... basically nothing? (Well, other than collecting paychecks for vaporware).

So "people made many mistakes" is a HELL of an understatement. These are supposed to be experts, but they clearly bit off far more than they could chew. Go look up "Dunning-Kruger effect" if you aren't already familiar with the term. EIGHT YEARS AGO they were *already four years late*, and their only visible achievement in all that time was to throw a massive temper tantrum over Mir (which had clearly started with none of those advantages AND gotten further in less time than that), which somehow led to Intel going back and sabotaging their driver specifically to not work with it. WTF?!

I have no idea how GOOD Mir actually was - but Canonical delivered it while the Wayland team was still just jerking around (something which continued for literally years afterwards), boasting about how great Wayland would be but not actually delivering anything at all. Since Mir was already more functional than Wayland at that time, and unlike Wayland was actually in a shippable state, if you have any experience with software at all it's almost impossible to imagine that Mir wouldn't have been at least as good as Wayland promises it will be someday, but would have BEEN at that point for at least 5 years already by now.

But Mir died (or was killed, if you prefer), so we'll never know. Which brings us back to where we came in: if Canonical had had enough money/influence/power to be able to stick with Mir, which they didn't at the time but potentially now do, the entire community would have benefited. Which is why this change in their financial situation is something that everyone who isn't a tribal fanboy should welcome.

Or at least, "welcome now that they've gotten over that CLA BS". That part absolutely deserved to DIAF, and I have no sympathy at all for THAT aspect of Mir. If not for that excess of greed, things might have been different, and X could ALREADY be a thing of the past - instead of us now being well over a DECADE into "Wayland will be ready next year".

[AHOY]

It's not a hard problem.

I was obviously talking about the entire feature set that Wayland is trying to (or have to indirectly, since is out of scope) replace. There are to this day protocols that people can't get a consensus on how they should look like, even if Wayland didn't exist. It's going beyond EWMH and things that were easy on X because you could do whatever the fuck you wanted in some sense. It's fine of course if you don't agree that it's a bad thing. There's also Flatpak, hardware features/constraints and new things to consider in the design. The entire ecosystem had to accept that Wayland was a thing and move to it, for reasons like having a different security model. Majority just waited to see if other people would do the job for them, Mir added even more fear and uncertainty. This would have happened no matter how the new display server looks like, if it wanted to fix long standing X issues.

I'm not defending Wayland technical merits, I'm sure there are many many flaws but even if it was designed like you wanted with a single implementation and centralized design it would probably have repeated errors from X and still take as many years to get mature. It would be missing many critical protocols that nobody knows to this day how it should work, but they are getting there.

But 13 years is a joke

We both know that's very disingenuous. It's not 13 years for reasons explained multiple times even in this thread. It's a good bait tho.

The rest of your post

As I suspected and as the rest of the comments showed me, Mir was nothing but a shallow attempt at gaining control. They couldn't win by force and certainly not by having superior software. They had nothing.

[arQon]

oof. There's a lot to unpack here...

I was obviously talking about the entire feature set that Wayland is trying to (or have to indirectly, since is out of scope) replace.

Wayland was supposed to be a replacement for X. Claiming that pieces are suddenly "out of scope" for some reason is just admitting that "We didn't design this thing properly and we're in over our heads". Did X suddenly add new features in the last 10 years that somehow made Wayland impractical? (Not a rhetorical question, but I'd be very surprised if the answer was "Yes", since, you know, "X is old and abandoned and" etc etc).

> There's also Flatpak

I'm sorry - how and on what planet does some random piece of userspace packaging affect the display server? I don't remember having to patch xserver every time I installed a new .deb.

> hardware features/constraints

You mean those things handled by the display driver? The things that Wayland hasn't needed to develop itself at all?

> and new things to consider in the design.

Sure, designs change over time, even if you're competent. But if your design has to consider so many "new things" that it takes 10x as long to deliver as it was supposed to, you're very much back at "out of our depth" again, to put it mildly.

> The entire ecosystem had to accept that Wayland was a thing and move to it, for reasons like having a different security model.

erm, wut? Almost NOTHING cares about the "different security model", even at the toolkit level. That's a red herring.

> Mir added even more fear and uncertainty.

From what I can see, the only "fear" around Mir (other than the CLA) was the Wayland team's fear that Mir was already the better option, and that if the whole of Red Hat didn't immediately jump to their defense and start slating it, they might lose their grip on the most fundamental piece of the display stack.

> I'm not defending Wayland technical merits, I'm sure there are many many flaws but even if (false claim of someone else's position snipped) it would probably have repeated errors from X and still take as many years to get mature.

Are you seriously trying to argue that because X was up and running in like a month but then slowly got extended over n decades, it's perfectly understandable that Wayland should be given the same n decades to become usable?!

"Second system syndrome" would be the next cliched anti-pattern for you to look up, I expect.

> but they are getting there.

Sure. Except for the parts that they still don't have an answer to, including some pretty basic functionality.

> We both know that's very disingenuous. It's not 13 years for reasons explained multiple times even in this thread. It's a good bait tho.

Oh? Please, enlighten me as to how it ISN'T 13 years, for "reasons" not covered in this thread at all, oh defensive one.
I guess that answers the "On what planet?" question from before though: one that apparently rotates around the sun much more slowly.

> Mir was nothing but a shallow attempt at gaining control.

And Wayland isn't?! You cannot POSSIBLY be so naive as to think that IBM isn't absolutely all about making sure they own every piece of the system that they can.
I mean, yeah, we're completely in agreement about the CLA being indefensible, but even pre-IBM, RedHat repeatedly leveraged its ownership of "open" projects to hurt its competition - as Canonical found out the hard way.

> They couldn't win by force and certainly not by having superior software. They had nothing.

You're not quite right there: "They DID win by force, and certainly not by having superior software. They had nothing." But the "they" there is the Wayland team, 8 years ago.

We're obviously not going to see eye to eye on this, but thanks for staying civil.

[AHOY]

oof indeed