Trying to run KDE 6 Plasma with Wayland....

[T-Aoki]

With the help of ChatGPT, I was able to get rid of this problem. It wrote patches for the plasma6-kwin port, and now if I start hopping between TTYs while the plasma wayland DE is running, I will be able to return to the DE's TTY and the DE will remain visible.

Unfortunately, from the view of interectual property, codes/patches generated by AIs has too many unclear zones to use publicly.
(Using only in home and single organization internals "could" be OK, though, but it's just MY current understanding and non official).

Using AI to determine where the root cause of the problem exists "could" be OK (again, just "MY current" understanding), but coding fixes should be better done purely by "actual human being" until the IP issues become clear and allowed by international law.

 

[astyle]

Unfortunately, from the view of interectual property, codes/patches generated by AIs has too many unclear zones to use publicly.
(Using only in home and single organization internals "could" be OK, though, but it's just MY current understanding and non official).

Using AI to determine where the root cause of the problem exists "could" be OK (again, just "MY current" understanding), but coding fixes should be better done purely by "actual human being" until the IP issues become clear and allowed by international law.

In that case, Open Source really reaps the benefits. If somebody, anybody solves a problem and submits a patch, even with help of ChatGPT, that actually counts as 'participation in Open Source'.

 

[T-Aoki]

In that case, Open Source really reaps the benefits. If somebody, anybody solves a problem and submits a patch, even with help of ChatGPT, that actually counts as 'participation in Open Source'.

First of all, this is just MY understanding and not at all legal conclusion.

Even if any softwares, such as FEM, statistical analysis, CAD/CAM and EDM are used for designs / reviewes / manufacturing of any products by natural human, the product is considered to be designed by the natural human.
So intellectual property (hereafter in this post, IP) goes to the natural human (often transferred to the organization the natural human belong to).
These IP includes patents, trademarks and copyrights and any others.

So why IP problem arises if the designs / implementations itself are done by natural human and use AI just as a helper software like above examples?
I think, in this case, IP clearly does NOT go to the AI and/or the organization that built the AI.

Of course, it's for "current state of AIs. Once AI advanced enough and considered as "artificial human" having basic human rights, the conclusion can be flipped. Maybe Rajendra in SF novel "Enemy is Pirate" by Chohei Kanbayashi could be an example.

 

[astyle]

First of all, this is just MY understanding and not at all legal conclusion.

Even if any softwares, such as FEM, statistical analysis, CAD/CAM and EDM are used for designs / reviewes / manufacturing of any products by natural human, the product is considered to be designed by the natural human.
So intellectual property (hereafter in this post, IP) goes to the natural human (often transferred to the organization the natural human belong to).
These IP includes patents, trademarks and copyrights and any others.

So why IP problem arises if the designs / implementations itself are done by natural human and use AI just as a helper software like above examples?
I think, in this case, IP clearly does NOT go to the AI and/or the organization that built the AI.

Of course, it's for "current state of AIs. Once AI advanced enough and considered as "artificial human" having basic human rights, the conclusion can be flipped. Maybe Rajendra in SF novel "Enemy is Pirate" by Chohei Kanbayashi could be an example.

Yeah, I get that this is just your understanding, T-Aoki...

It's just that I sometimes wonder, where should the line be drawn? Just how far can one take the very idea of 'Intellectual Property' before it becomes just flat-out ridiculous on its face and impossible to defend?

I actually have a separate thread (Thread licensing-rant-debate-thread.90051) that touches on this.

I personally think that since AI is just a tool, credit for the work should go to the user of the tool, rather than the creator. I mean, if I use a Komatsu backhoe to dig a trench to drain the water, that's MY efforts, the manufacturer has no business sticking his nose into my business. Komatsu ain't Ferrari.

 

[T-Aoki]

Yeah, I get that this is just your understanding, T-Aoki...

It's just that I sometimes wonder, where should the line be drawn? Just how far can one take the very idea of 'Intellectual Property' before it becomes just flat-out ridiculous on its face and impossible to defend?

I actually have a separate thread (Thread licensing-rant-debate-thread.90051) that touches on this.

I personally think that since AI is just a tool, credit for the work should go to the user of the tool, rather than the creator. I mean, if I use a Komatsu backhoe to dig a trench to drain the water, that's MY efforts, the manufacturer has no business sticking his nose into my business. Komatsu ain't Ferrari.
View attachment 23439

I think that the difference between backhoe and AI is that "does the operator exactly and clearly understanding what is going on or not".

With the example of backhoe:

• Legacy developement would be skilled and experience operator directly operate the backhoe.
• Using AI to determine the root cause of the problem would be newbie operator operates AI-supported backhoe that assists somewhere extra skills are required. The operator still knows and understands where and how to dig/entiten/flatten.
• Using AI for coding itself would be design center of the whole construction site designates purely-AI-operated and unmanned backhoe to do the whole operations. Yes, still the design is possibly done by natural human engineers, but what about backhoe operator?

Video of AI-operated backhoes. There, also introduced "remote control by natural human, too.

Couldn't find any movie for "AI assisted backhoe" other than safety assists.
I've saw one on Japanese television.

 

[vmisev]

It's just that I sometimes wonder, where should the line be drawn? Just how far can one take the very idea of 'Intellectual Property' before it becomes just flat-out ridiculous on its face and impossible to defend?
View attachment 23439

Completely off topic, but talking about 'Intellectual Property' I’m only surprised that Intel didn’t sue Komatsu because of that ‘Utility’ logo. Heck, Intel sued eFing Yoga studio for “Yoga inside” slogan. I got curious, so I dug a bit, and turns out that Komatsu did apply for that logo trademark registration, but “The trademark application was registered, but subsequently it was cancelled or invalidated and removed from the registry” /q

 

[T-Aoki]

Completely off topic, but talking about 'Intellectual Property' I’m only surprised that Intel didn’t sue Komatsu because of that ‘Utility’ logo. Heck, Intel sued eFing Yoga studio for “Yoga inside” slogan. I got curious, so I dug a bit, and turns out that Komatsu did apply for that logo trademark registration, but “The trademark application was registered, but subsequently it was cancelled or invalidated and removed from the registry” /q

This would be my final post about IP unrelated with KDE6 in this thread.
Already too off-topic.

In Komatsu case, looking into the page you linked, it was just "not renewed" at Jan.10, 2019 with regard to its "Document" section.
And trademarks usually strongly tied to "actual business". I "think" Intel cannot claim the trademark to be rejected maybe except for Primary US Class 021 and possibly 034. And International Class 007 doesn't seem to related with Intel.

And what's confusing is that in the page your link points to, in Trademark Filing History section, the reason not renewed is "DEATH OF INTERNATIONAL REGISTRATION", but it seems to be active for Komatsu Japan as Reg.No. 969969 by searching here with keywords "Komatsu Utility".

If you want more discussion, please DM me, not responding here.

 

[DanielBowman]

Unfortunately, from the view of interectual property, codes/patches generated by AIs has too many unclear zones to use publicly.
(Using only in home and single organization internals "could" be OK, though, but it's just MY current understanding and non official).

Using AI to determine where the root cause of the problem exists "could" be OK (again, just "MY current" understanding), but coding fixes should be better done purely by "actual human being" until the IP issues become clear and allowed by international law.

International Law is the Law of the Stronger (first lecture in my study course from prominent Czech scholar).
Laws are letters to Santa Clause (anonymous scholar).
It all depends on the political will to prosecute or not i.e. Epstein clients and handlers vs 5 years for facebook posts by some nobody.
Of course down the road some Samuel Altman's cousin lawyer will try to sue you for billions, because ChatGPT wrote some BS code for your successful project.

 

[astyle]

International Law is the Law of the Stronger (first lecture in my study course from prominent Czech scholar).
Laws are letters to Santa Clause (anonymous scholar).
It all depends on the political will to prosecute or not i.e. Epstein clients and handlers vs 5 years for facebook posts by some nobody.
Of course down the road some Samuel Altman's cousin lawyer will try to sue you for billions, because ChatGPT wrote some BS code for your successful project.

Please do not get off-topic in this thread. This thread is for posting progress towards Plasma Wayland being functional on FreeBSD.

 

[tuaris]

Supposedly this is doable now: https://euroquis.nl/kde/2025/09/07/wayland.html
But so long as Wayland doesn't work in VMWare on FreeBSD (https://github.com/freebsd/drm-kmod/issues/33), it won't work for me.

I sure hope someone doesn't forget about us VMWare users. I can't stand Virtualbox.

 

[astyle]

Seems like the guy at codeberg still has issues with SDDM and CTRL-C in Wayland. I'm a bit surprised, SDDM worked fine for me for login, but I haven't had a chance to play with CTRL-C, and now that I look back on what I did, the notes I took, I'm surprised at myself that I didn't play with CTRL-C earlier

I sure hope someone doesn't forget about us VMWare users. I can't stand Virtualbox.

ouch... VirtualBox is at v. 7 in Ports right now... what's your beef with VBox, anyway?

 

[tuaris]

ouch... VirtualBox is at v. 7 in Ports right now... what's your beef with VBox, anyway?

I've answered that question on these forums already , basically I just don't find it to be as polished as VMWare.

 

[Espionage724]

Seems like the guy at codeberg still has issues with SDDM and CTRL-C in Wayland. I'm a bit surprised, SDDM worked fine for me for login, but I haven't had a chance to play with CTRL-C, and now that I look back on what I did, the notes I took, I'm surprised at myself that I didn't play with CTRL-C earlier

SDDM the log-in screen? I looked at it for a minute but can't figure out why I'd ever press Ctrl + C there (I type a pass and Enter; GNOME an extra Enter)

what's your beef with VBox, anyway?

Not sure about the other person, but I had lower FPS with FlightGear in VirtualBox vs VMWare Player.

Years ago I think VMware had a better or faster solution to 3D acceleration in Windows guests. Iirc I preferred Fusion on macOS over Parallels too (maybe seamless windows?)

 

[astyle]

SDDM the log-in screen? I looked at it for a minute but can't figure out why I'd ever press Ctrl + C there (I type a pass and Enter; GNOME an extra Enter)

Well, the DM part of SDDM refers to Display Manager. And yes, that does include turning the monitor on and off, believe it or not. Not a very logical way to combine authentication with power management duties, would be nice to have a cleaner separation, but we got what we got.

My thinking goes, I'm just hoping that when I copy text in Kate in a KDE Wayland session (using CTRL-C), the whole session doesn't crash. Because if it does - that will make for a very unfortunate showstopper bug. And who knows how easy it will be to fix. Sometimes a visible bug is merely a symptom of an erroneous implementation of something else altogether. As an example, earlier in this thread, there was talk of a right-click bug that was solved with a compilation flag that needed to be set for kwin6... Completely un-intuitive unless you have a very good handle on the source code.

 

[LibreQuest]

Plasma Wayland is working on my RX 6400 computer.

 

[astyle]

Plasma Wayland is working on my RX 6400 computer.

Screenshots, please, complete with QT/KF6/Plasma6 versions, and a note on whether SDDM works, and whether screen timeouts / power management are functional. Oh, and if Ctrl-C is not a problem in Kate, that would be nice to note, too.

There's still a lot to be ironed out, so the more detail, the better.

 

[LibreQuest]

CTRL-C crashes KATE. SDDM works to login to wayland. Plasma 6.4.5. I am able to block and unblock power management. I'm on latest for 14.3 so whatever KF6 and QT6 versions are latest is what I have installed. I didn't really do anything special. I just installed seatd, enabled in rc.conf and logged in with SDDM.

 

[astyle]

CTRL-C crashes KATE. SDDM works to login to wayland. Plasma 6.4.5. I am able to block and unblock power management. I'm on latest for 14.3 so whatever KF6 and QT6 versions are latest is what I have installed. I didn't really do anything special. I just installed seatd, enabled in rc.conf and logged in with SDDM.

This is awesome! Seriously!

But yeah, Ctrl-C crashing Kate - that is a serious showstopper, so I really appreciate that we got that out in the open for this thread.

At some point I'll be curious to see if the Linux camp still suffers from this bug or not, but if they actually fixed that Ctrl-C bug, that's a hint for the FreeBSD camp, I'd think.

I personally am just waiting for 15-RELEASE to be out in December before making the switch to Plasma Wayland...

One more question: Are you using packages or ports?

 

[LibreQuest]

This is awesome! Seriously!

But yeah, Ctrl-C crashing Kate - that is a serious showstopper, so I really appreciate that we got that out in the open for this thread.

At some point I'll be curious to see if the Linux camp still suffers from this bug or not, but if they actually fixed that Ctrl-C bug, that's a hint for the FreeBSD camp, I'd think.

I personally am just waiting for 15-RELEASE to be out in December before making the switch to Plasma Wayland...

One more question: Are you using packages or ports?

Packages. Just whatever is available in latest is what I am using. But yeah, gaming with Wayland is smooth and perfection.

EDIT: Yeah, I will just use wayland for games if I'm playing games or watching movies. I like how smooth Wayland is for video. But KATE crashing is crazy.

 

[astyle]

I do think that it's important to note that it's a problem with the Ctrl-C key combination within the Wayland session, rather than Kate itself. Kate is merely a good way to demonstrate the presence of the bug. I think that any program where Ctrl-C is supposed to work will crash in a Plasma Wayland session.

 

[vmisev]

Just a guess (I'm not using Wayland, only X), but can you guys try to comment out sddm and start plasma from vt with dbus-run-session startplasma-wayland (or with wthatewer is a proper way to start startplasma-wayland) and the see if ^C works as it should? It's maybe more of the fight with sddm and Wayland, than Plasma and Wayland. As I said, just a guess...

 

[astyle]

That's one thing to try (with the latest stuff installed). With a divide-and-conquer approach to troubleshooting, it's a way to narrow things down. Just definitely knowing that yeah, not where the problem is - that reduces the answer space.

 

[vmisev]

Okay, just did some small test in VBox FreeBSD VM - sddm is commented out anyway - login is into vt, added seatd, sudo service seatd onestart, commented out in ~/.xinitrc my startplasma-x11 line, added exec dbus-launch --exit-with-x11 ck-launch-session /usr/local/bin/startplasma-wayland, startx and although I see only small 1024x768 window in the middle of larger desktop (native 1920x10800), ^C is not causing any problems in konsole, and works normally in kate as 'copy' ( ^V pastes normally as well). kinfocenter is showing that Wayland is running. Hope this will help to narrow it down.

 

[astyle]

added exec dbus-launch --exit-with-x11 ck-launch-session /usr/local/bin/startplasma-wayland, startx

That startx part is kind of the problem with your setup... the whole thing is supposed to work without Xorg at all, even in a VM. Using the .xinitrc is cheating, sorry. The setup needs to work as indicated on this page: https://euroquis.nl/kde/2025/09/07/wayland.html . This late in the game, having reproducible setups becomes actually important when hunting down showstoppers.

 

[vmisev]

That startx part is kind of the problem with your setup... the whole thing is supposed to work without Xorg at all, even in a VM. Using the .xinitrc is cheating, sorry. The setup needs to work as indicated on this page: https://euroquis.nl/kde/2025/09/07/wayland.html . This late in the game, having reproducible setups becomes actually important when hunting down showstoppers.

That was only way that I could start Wayland in VBox, script from the link you proposed doesn't work because kwin_wayland_drm can't open drm device /dev/dri/card0 - which doesn't exist in VBox guest. I have no intention to pollute nor screw my on the metal install – I have 0 interest in running Wayland, just wanted to help. I still think that ^C issue is sddm/ Wayland problem, good luck with your investigation.