Running Wayland: It Works, But A Lot Of Work Remains

[kazetsukai]

Another example, we had compiz 0.8 written in C. The developer rewritten it for C++. Of course, for a couple of versions it's shit (12.04 being a shitty standout). What was so wrong about keeping it in C and improving it from there? It was 95% ok, but nooooo we have to start again...

...You're not a developer, are you (then why the hell are you questioning their design decisions?) Sometimes, the rats nest gets too deep, the spaghetti too entwined, the hair too tangled to really sort everything out, and your project SCREAMS for simplification. This is when a developer cuts his losses, looks hard at the lessons he has learned and starts anew.

Linux could take over a larger marketshare, but the devs jump to do a thing that already worked from scratch, they get all the media attention(ohhh look wayland, I'm cumming!), and it's never good enough for the real market.

This problem isn't really what is holding Linux back. When KDE 4 rolled around, KDE3 still existed. Gnome2 still exists. Currently well supported software still exists in the Linux world, and that software is often getting bugfixes for a while. Use cases have changed, hardware has advanced, and X has needed this refresh for at least 10 years now.

And when they'll release it we won't see an improvement, in fact most likely we will have problems...and then we will start from scratch once again, this time we'll do it right!

When I started using KDE4 it was the earliest 4.0 releases. Lets give you a little context: I used 5 or 6 monitors at the time, and KDE4 didn't support TWO X screens, let alone 6, until well into KDE 4.5. I was hanging in there for years, banging on the door, submitting bugs, etc.... it drove me NUTS. But, meanwhile, somewhere else in the world... someone was still using KDE 3 the entire time, and didn't have those problems or the frustration. They happily used KDE 3 beyond its EOL. I bet there are people that STILL use KDE 3 out there.

You don't have to upgrade. Its a choice, one you make based on educated decisions (hopefully). If you're just going bleeding edge, ALL THE TIME, prepare for the problems. For everyone else, there's LTS distros.

[johnc]

If you're just going bleeding edge, ALL THE TIME, prepare for the problems. For everyone else, there's LTS distros.

Isn't Ubuntu 12.04 an LTS release?

[kazetsukai]

Isn't Ubuntu 12.04 an LTS release?

There's also 8.04 and 10.04, which are, still supported.

[johnc]

There's also 8.04 and 10.04, which are, still supported.

Yes but the point is someone who says, "Let me try out this Ubuntu stuff" will download 12.04, install it, experience an absolute catastrophe of a disaster, and then scratch Linux off the list for the next 10 years.

[Ancurio]

I am not saying you personally try to get people on linux. But I don't think you are against the idea of linux having a larger marketshare (think more developers involved -> maybe less bugs??). But it's a chicken and egg problem: to have a larger marketshare you need less bugs in the first place and to be more user friendly. So, instead of having a system that is 95% good and throwing it away to start from scratch (tell me how many times has this happened in linux? Everybody starts from scratch and thinks this new design is the shit while the previous ones are shit) like it's happening now with x and wayland, I say fuck wayland just improve x.

Another example, we had compiz 0.8 written in C. The developer rewritten it for C++. Of course, for a couple of versions it's shit (12.04 being a shitty standout). What was so wrong about keeping it in C and improving it from there? It was 95% ok, but nooooo we have to start again...and then again..and then again. Linux could take over a larger marketshare, but the devs jump to do a thing that already worked from scratch, they get all the media attention(ohhh look wayland, I'm cumming!), and it's never good enough for the real market. And when they'll release it we won't see an improvement, in fact most likely we will have problems...and then we will start from scratch once again, this time we'll do it right!

Have you ever seriously worked on any big project? I don't want to make any assumptions, but you sound like you have no developer experience.

Let me give you a real world example: in a lot of cases, when a new house is to be built replacing an old one, the old one isn't restored, it's smashed into the ground.
Now why is that? It's for a very simple reason: smashing and starting completely a new is often times WAY cheaper than restoring.
(And in software you don't even have to smash the old one, see below)
Imagine you have a wood house, but due to circumstances, really need it to be made out of cement instead. That's not something you can
just "improve" on the old house, is it?

You might think that X and Wayland are very similar simply for the fact that they serve a similar role, but if you made any serious attempt at comprehending them
you would soon realize that they are fundamentally different approaches to the problem. Hoegsberg himself is a senior contributor to X (I think),
so he must know very well that yet another extension to X won't do the trick.
I'd suggest you to head over to wayland's website and read the comparison to X give there, it's pretty insightful.

There's two major advantages I see to starting from scratch:
1. Obviously, you don't have to worry about cleaning up bad/old code. This frees up a lot of time for other things.
2. You leave the old project intact. Did you ever consider all the people heavily relying on X's features and behavior?
X itself isn't fundamentally bad, it's still good for many use cases (ie. remote sessions), and as many said already, will continue to live on for a very long time.

Oftentimes, people confuse things like in the case with PulseAudio, calling it an "ALSA replacement" when it's really not
(It's a proper solution for the problem dmix was a quick hack solution to).

[kazetsukai]

Yes but the point is someone who says, "Let me try out this Ubuntu stuff" will download 12.04, install it, experience an absolute catastrophe of a disaster, and then scratch Linux off the list for the next 10 years.

You do know that Wayland didn't make the cut for 12.04, right? Not surprising, considering Canonical's (lack of) contributions....

Each release, you get the latest stable software stack, and it receives bugfixes until the disto EOLs. So if you download 12.04, you get 12.04. If you download 10.04, you get 10.04. As far as I know, there's no reason for normal users to be recommended 10.04 or 8.04 over 12.04.

[Ancurio]

Yes but the point is someone who says, "Let me try out this Ubuntu stuff" will download 12.04, install it, experience an absolute catastrophe of a disaster, and then scratch Linux off the list for the next 10 years.

Then the fault lies entirely with Ubuntu and not the software developers. They don't dictate what ships with what.

[Akka]

I am not saying you personally try to get people on linux. But I don't think you are against the idea of linux having a larger marketshare (think more developers involved -> maybe less bugs??). But it's a chicken and egg problem: to have a larger marketshare you need less bugs in the first place and to be more user friendly. So, instead of having a system that is 95% good and throwing it away to start from scratch (tell me how many times has this happened in linux? Everybody starts from scratch and thinks this new design is the shit while the previous ones are shit) like it's happening now with x and wayland, I say fuck wayland just improve x.

Another example, we had compiz 0.8 written in C. The developer rewritten it for C++. Of course, for a couple of versions it's shit (12.04 being a shitty standout). What was so wrong about keeping it in C and improving it from there? It was 95% ok, but nooooo we have to start again...and then again..and then again. Linux could take over a larger marketshare, but the devs jump to do a thing that already worked from scratch, they get all the media attention(ohhh look wayland, I'm cumming!), and it's never good enough for the real market. And when they'll release it we won't see an improvement, in fact most likely we will have problems...and then we will start from scratch once again, this time we'll do it right!

The new c++ version of compiz is not considered stable, at least if you not use the default unity set up. According to upstream you should use the 0.8.8 version if you run outside this user-case as I understand it. I get the impression that most of the distributions who is shipping compiz ship 0.8 series.++

[BO$$]

Have you ever seriously worked on any big project? I don't want to make any assumptions, but you sound like you have no developer experience.

Let me give you a real world example: in a lot of cases, when a new house is to be built replacing an old one, the old one isn't restored, it's smashed into the ground.
Now why is that? It's for a very simple reason: smashing and starting completely a new is often times WAY cheaper than restoring.
(And in software you don't even have to smash the old one, see below)
Imagine you have a wood house, but due to circumstances, really need it to be made out of cement instead. That's not something you can
just "improve" on the old house, is it?

You might think that X and Wayland are very similar simply for the fact that they serve a similar role, but if you made any serious attempt at comprehending them
you would soon realize that they are fundamentally different approaches to the problem. Hoegsberg himself is a senior contributor to X (I think),
so he must know very well that yet another extension to X won't do the trick.
I'd suggest you to head over to wayland's website and read the comparison to X give there, it's pretty insightful.

There's two major advantages I see to starting from scratch:
1. Obviously, you don't have to worry about cleaning up bad/old code. This frees up a lot of time for other things.
2. You leave the old project intact. Did you ever consider all the people heavily relying on X's features and behavior?
X itself isn't fundamentally bad, it's still good for many use cases (ie. remote sessions), and as many said already, will continue to live on for a very long time.

Oftentimes, people confuse things like in the case with PulseAudio, calling it an "ALSA replacement" when it's really not
(It's a proper solution for the problem dmix was a quick hack solution to).

In fact you sound like you have no experience : http://www.joelonsoftware.com/articl...000000069.html

Never rewrite from scratch my friend...never. All that code you consider crap actually has a lot of bugfixes, the new code has a new array of problems and most likely will also repeat some bugs from the past....

[BO$$]

The new c++ version of compiz is not considered stable, at least if you not use the default unity set up. According to upstream you should use the 0.8.8 version if you run outside this user-case as I understand it. I get the impression that most of the distributions who is shipping compiz ship 0.8 series.++

Ubuntu 12.04 that will get on more computers that usual if we listen to Mr. Shuttleworth uses 0.9.something. And 11.10 also used 0.9.something if I'm not mistaken.