We Have Mir & Wayland, But There Still Could Be X12

[erendorn]

I have been using X for remote displays quite satisfactorily for my touchscreen applications for over a decade and find nothing really wrong with it. NX does improve performance a bit but there's nothing broken in X that prevents or even hinders my use of it. Since Wayland removes the network transparent capabilities of X it is entirely unsuitable as a reworking of X.

I'm sorry if I appeared rude. I have read again the thread and indeed you jumped in the discussion only defending the fact that X11 network remote transparency is still used/can be made to work, and it's a valid point.

But by saying that wayland cannot work for you since it removes the network transparent capabilities of X, you are arguing that wayland should be network transparent. And that is not true.

In the modern Linux graphic stack (not limited to wayland), the display server displays (and only read buffers), and the toolkits render, using whatever means and acceleration and API and ABI they like, including through the network if they fancy.
The display server part can do remote displaying like VNC, and I think it has already been demonstrated for Weston with a VNC backend.
Any toolkit can do remote rendering like X11, and it may even have already been done using xlib (server side) and xwayland (client side).

There isn't yet seamless remote rendering in wayland. That's expected because there isn't yet seamless wayland anyway.
But it can be done, and it can be done without wayland having network transparency.

In short, it is not an issue in the design of the modern Linux graphic stack, it is an issue with the fact that this new graphic stack and ecosystem is not feature complete yet (far from it).

[pq__]

It is a conflict of interest, but I do not care enough about Wayland to track its development to catch these things. The closest of which I am aware is an incident where Intel reverted a commit to support Canonical's Mir from their driver; it had the side effect of pushing people toward the technologies that they are using in Tizen. Clearly, they have a bias. Anyway, the converse question is do you know of any cases where Wayland's development has compromised Tizen's development goals. I cannot imagine an executive at Intel letting that happen (at least not without firing people).

You seem pretty confident in believing that there actually can be any permanent conflicts.

Wayland is a protocol based heavily on extensions. The core is very small and fairly universal, and even that is divided into parts that might be left unimplemented in some circumstances. When conflicting cases appear, they are separated into different extensions, so everyone can live happily together. If upstream does not approve of some extension, that extension can perfectly well live outside of Wayland upstream, and everyone willing can use it. There is no way to really prevent that, either. It is up to individual toolkits (Wayland clients) and compositors (Wayland servers) to decide which parts they implement and rely on.

Whether some piece of software (as opposed to protocol) that needs active maintenance supports some particular features is a completely different matter.

On another note, patches written by people with an Intel email address have been rejected.

[Remote User]

I was reading the X12 list back in 2010. It is essentially a list of what has been done wrongly in X11. New protocol requires applications to be updated. So it's better to do with something like Wayland, which is a more radical redesign (and follows the philosophy of leaving much stuff to clients), suited to modern graphic hardware, while the network transparency requirement is removed as this is anyway semi-broken today (you anyway have to use something like NX - which does many tricks to shortcut X protocol issues - to really get useful remote desktop performance).

I have been using X for remote displays quite satisfactorily for my touchscreen applications for over a decade and find nothing really wrong with it. NX does improve performance a bit but there's nothing broken in X that prevents or even hinders my use of it. Since Wayland removes the network transparent capabilities of X it is entirely unsuitable as a reworking of X.

[smorovic]

I was reading the X12 list back in 2010. It is essentially a list of what has been done wrongly in X11. New protocol requires applications to be updated. So it's better to do with something like Wayland, which is a more radical redesign (and follows the philosophy of leaving much stuff to clients), suited to modern graphic hardware, while the network transparency requirement is removed as this is anyway semi-broken today (you anyway have to use something like NX - which does many tricks to shortcut X protocol issues - to really get useful remote desktop performance).

[Remote User]

But then you can just keep X11 if it works for you. What you can't, is arguing that wayland should be network transparent.

I have not argued that Wayland should be network transparent. I don't really understand why you have to take a shitty attitude toward anything I've written, put words in my mouth, and misrepresent what I have written here.

[erendorn]

It not only doesn't solve any problems I have - it also destroys what works for me. X works perfectly for what I do with touchscreen GUIs at customer locations all around the world. I will never understand why some of you just can't deal with the fact that X makes a lot of things not only possible, but also easy, which would otherwise simply not be possible.

But then you can just keep X11 if it works for you. What you can't, is arguing that wayland should be network transparent.

[mannerov]

I'm rooting for wayland. I want wayland to succeed to the point where i'd be willing to donate to help it's development if i could actually try it and see if i like it, but its just not in any even remotely usable state yet from what i hear...

But Wayland's development is too slow and Mir is just some ubuntu junkpile that i wont ever want to touch, if X12 would be a X server written from scratch with modern hardware only in mind (forgetting everything about support for stuff from pre-2000) then i might just change my mind and support X12 instead.

It's becoming pretty usable.

Only a few details needs to be clarified, and some bugs to solve.
If you want to test last Weston, you can try to install it on an external hard drive with Arch Linux, or try Fedora 20 (less complete Wayland implementation however).

If you are a programmer, you can help by solving bugs.

[rabcor]

I'm rooting for wayland. I want wayland to succeed to the point where i'd be willing to donate to help it's development if i could actually try it and see if i like it, but its just not in any even remotely usable state yet from what i hear...

But Wayland's development is too slow and Mir is just some ubuntu junkpile that i wont ever want to touch, if X12 would be a X server written from scratch with modern hardware only in mind (forgetting everything about support for stuff from pre-2000) then i might just change my mind and support X12 instead.

[ryao]

Thanks for that, I'll check it out.

I edited my post to say the following around the time you posted this:

Edit: It turns out that it was either the placebo effect or the effect of a freshly started Xorg server. That setting does not persist across Xorg sessions. :/

Anyway, if you find a setting that helps (that for some reason is not the default), please let me know. I would be interested,

On a related note, I noticed that 2D performance from the Nvidia blob on Solaris seems qualitatively better than 2D performance from the Nvidia blob on Linux. I am not sure why that is. It might be a GNOME 2.32 versus KDE 4.11 difference.

[ryao]

So, you just make unfounded assumptions about how Wayland development is being directed?

They are founded on the belief that no company would play such a role in software without ensuring that it works well with their other stuff first. It is a rather sensible thing to believe. The only company of which I am aware that failed to do this was Sun Microsystems with the Lustre Filesystem.