X.Org Server 1.20.11 Released Due To New Security Advisory

[reavertm]

Wayland is exactly that, a better X11 breaking backwards compatibility where it is necessary to address its shortcomings.

Except without full/best/reference/production implementation. And last time I checked it was primary problem of Wayland - everyone needs to implement its protocols. Reference implementation is just for demonstration purpose apparently, a proof of concept. You did not address that part so my question remains open.

[CochainComplex]

tsk! that's nothing, nowadays everybody accepted the end of X11.
the real interesting comments comes when the article is about systemd or btrfs

Well, X11 vs Wayland really calmed down a bit lately. zfs vs btrfs is a good one. Best is still e.g. "nvidia releases XYZ opensource framework"

[Alexmitter]

I will ask stupid question maybe.

Is Wayland (releasing bunch of protocols) really the best in class design to realise display server / drivers framework in Open Source environment?
Windows has been benefitting from WDF or WDM over the years and in general from being proprietary and fully under control of single entity, having de-facto or fully production ready reference implementation.

As was X11 somewhat. Wouldn't it be better to just have better X11, breaking backward compatibility when necessary to address its shortcomings?

Wayland is exactly that, a better X11 breaking backwards compatibility where it is necessary to address its shortcomings.

[reavertm]

I will ask stupid question maybe.

Is Wayland (releasing bunch of protocols) really the best in class design to realise display server / drivers framework in Open Source environment?
Windows has been benefitting from WDF or WDM over the years and in general from being proprietary and fully under control of single entity, having de-facto or fully production ready reference implementation.

As was X11 somewhat. Wouldn't it be better to just have better X11, breaking backward compatibility when necessary to address its shortcomings?

[dpeterc]

No, I know someone who does not. In his eyes, Xorg is way better because reasons. And Arcan is way better too. And then he send me a issue on the wayland repo where someone has the following scenario:
* He starts client via the terminal
* He suspends the client (technically freezing the process)
* The compositor at some point has a timeout and closes the connection to the client

Then he told me "Imagine if you have a bitcoin wallet and it fetches gigabytes of data from the harddisk and your compositor kills it" and I did die internally from reading that.

People, if your program stalls main() for a large amount of time to fetch data from the harddrive, you do something massively wrong and should not be anywhere near to a computer.

It is not a matter or large or small amount of data. I have a GUI program which reads a file, and it can not continue until it is fully read. It can be 10 bytes, it can be megabyte. Files can be on a local or on remote network server. Program does not know. I have a wired connection. If I accidentally trip on the wire and unplug the network, my process is stalled until i reattach it. When I do, it continues undisturbed. Just what I want. My program is stuck in fread() until it succeeds or fails. Now call me stupid, but this is how practically all programs work.

Program can be swapped out from RAM to disk, and until it is read back in, it can not continue. It may run slower, but it will work.

[Alexmitter]

tsk! that's nothing, nowadays everybody accepted the end of X11.

No, I know someone who does not. In his eyes, Xorg is way better because reasons. And Arcan is way better too. And then he send me a issue on the wayland repo where someone has the following scenario:
* He starts client via the terminal
* He suspends the client (technically freezing the process)
* The compositor at some point has a timeout and closes the connection to the client

Then he told me "Imagine if you have a bitcoin wallet and it fetches gigabytes of data from the harddisk and your compositor kills it" and I did die internally from reading that.

People, if your program stalls main() for a large amount of time to fetch data from the harddrive, you do something massively wrong and should not be anywhere near to a computer.

[Alexmitter]

Dead.. when both KDE/GNOME compositors *both* can't turn on my DisplayPort on this laptop... Wayland is not ready, still. I can't switch to it at all.

[ 126.282206] amdgpu 0000:01:00.0: amdgpu: SRBM_SOFT_RESET=0x00100040
[ 126.449983] [drm] Got external EDID base block and 2 extensions from "edid/Samsung-LC27G7xT.edid.bin" for connector "DP-2"

Wayland is a protocol between clients(programs you open) and the compositor. It has nothing to do with your Samsung display having as broken edid, in fact it has nothing to do with Displays at all, drm/dri/kms do that. KDE and Gnome parse the edid info in a similar way and it is more to specification then the way Xorg parses it, and that is likely the reason why Xorg accepts your display.

[cynic]

I love the x11 or wayland articles....popcorn and reading the comment section

tsk! that's nothing, nowadays everybody accepted the end of X11.
the real interesting comments comes when the article is about systemd or btrfs

[fafreeman]

Thanks for the response. I did some digging and you are correct. Unfortunately SDDM has a hard requirement for the xorg-x11-server. I hope that is some thing they intend to correct before the final release of 34. It would be odd to have a release whose main feature is to remove X but then force you to take it anyway over the greeter which is going to send you to Wayland. I'm sticking with the idea that the best way to update X is to remove it.

running gnome 40 myself, gdm is the same too. which i find odd as gdm uses wayland by default on everything except nvidia.

[angrypie]

Dead.. when both KDE/GNOME compositors *both* can't turn on my DisplayPort on this laptop... Wayland is not ready, still. I can't switch to it at all.

[ 126.282206] amdgpu 0000:01:00.0: amdgpu: SRBM_SOFT_RESET=0x00100040
[ 126.449983] [drm] Got external EDID base block and 2 extensions from "edid/Samsung-LC27G7xT.edid.bin" for connector "DP-2"

I'm not sure it's the Wayland compositors for KDE/GNOME or if its amdgpu.ko thats broken but this works fine in X...

If nobody is going to release 1.21, maybe it's time someone took the reins of this, if its documented how the releases are made. I would if I had time...

Maybe--just maybe--reporting a bug and giving feedback on your bug is easier than maintaining Xorg by yourself.