Solved Is it possible to upgrade plasma5 -> plasma6?

I am running 14.2p2 on the latest repo on my laptop, a Thinkpad W520 with the i915 driver. I initially attempted to upgrade my laptop and desktop (nvidia driver, also 14.2p2 on the latest repo) from plasma5 to plasma6, about the time that pkg 2.0.0 came out. I was able to roll back to a previous boot environment, and have attempted the upgrade several times. I followed the UPGRADING document, (pkg remove -f plasma5-plasma && pkg install plasma6-plasma; pkg autoremove, followed by a pkg upgrade and reboot).

Each time, the upgrade completes, but I am unable to get KDE to start. I try both the X11 and the wayland versions, and neither will start. With the X11 instance, I get a black screen with a mouse cursor. With plasma-wayland, I get a blank screen and no mouse cursor...

Another thing I noticed is that from SDDM (which reconfigured itself from my settings from before the upgrade), I no longer have access to any of the virtual terminals. So VT1-VT8 do not seem to exist any longer.

So is it possible to get KDE upgraded, preferably without reinstalling my system? Can anyone give me a point in the right direction?

Thanks,
--vr
 
do you have D-Bus enabled in your /etc/rc.conf ?
Enabled dbus as suggested, to the "before upgrade" bootenv, did another upgrade:
  • pkg update
  • pkg install pkg
  • pkg remove -f plasma5-plasma && pkg install plasma6-plasma
  • pkg autoremove
  • pkg upgrade
  • ...and then rebooted. System restarted, through the FreeBSD loader, started into the boot, then the screen blanked and stayed that way. It never got to the sddm boot screen, nor was it able to go to the virtual terminals.
 
do you have D-Bus enabled in your /etc/rc.conf ?
Okay, I did manage to get it to the sddm login prompt. And with the dbus_enable=YES in rc.conf, I still get a black screen when I log in to Plasma + wayland, and a black screen with a mouse cursor that does nothing but moves around the screen when selecting Plasma + x11.

It really should not be this difficult to do this upgrade.
 
I posted this a while back.
And there's this:
 
I posted this a while back.
And there's this:
But these are ports-specific from my reading of the situation. My understanding is that you are not supposed to cross the streams between ports and pkgs...
 
But these are ports-specific from my reading of the situation. My understanding is that you are not supposed to cross the streams between ports and pkgs...
Packages are built from ports. I've never heard of a ports-specific thing that doesn't continue on into a package eventually but one could enlighten me if there is such a thing.
 
Packages are built from ports. I've never heard of a ports-specific thing that doesn't continue on into a package eventually but one could enlighten me if there is such a thing.
It's actually, to my understanding, less about packages themselves, and more about the cadence of the package releases. By definition, there will be a lag between the release of a package in pkg, as opposed to the ones in ports. Ports will normally be ahead of pkg. So I have read that, especially with complex packages, you want to stick to one or the other.

I used to encounter a similar situation running Debian, which is why I generally ran unstable. Because the testing branch would do something like having one library that would not get updated, and then your whole desktop or whatever would end up getting deinstalled or stop working.
 
Ports will normally be ahead of pkg.
Always. ;) As official pkgs are built from ports with "default options".
So if anyone building ports with non-default option (and/or non-blank and affecting /etc/make.conf) would have different behavior between official pkg and locally built ones.
 
I am running 14.2p2 on the latest repo on my laptop, a Thinkpad W520 with the i915 driver. I initially attempted to upgrade my laptop and desktop (nvidia driver, also 14.2p2 on the latest repo) from plasma5 to plasma6, about the time that pkg 2.0.0 came out. I was able to roll back to a previous boot environment, and have attempted the upgrade several times. I followed the UPGRADING document, (pkg remove -f plasma5-plasma && pkg install plasma6-plasma; pkg autoremove, followed by a pkg upgrade and reboot).

Each time, the upgrade completes, but I am unable to get KDE to start. I try both the X11 and the wayland versions, and neither will start. With the X11 instance, I get a black screen with a mouse cursor. With plasma-wayland, I get a blank screen and no mouse cursor...

Another thing I noticed is that from SDDM (which reconfigured itself from my settings from before the upgrade), I no longer have access to any of the virtual terminals. So VT1-VT8 do not seem to exist any longer.

So is it possible to get KDE upgraded, preferably without reinstalling my system? Can anyone give me a point in the right direction?

Thanks,
--vr
What else is relevant? yesterday, KDE 6 became the main version of KDE. Just upgrade the ports (pkg install kde). most likely, this will be enough. just keep in mind that there is a chance to disable the system.

And I forgot to say this: We need to update the system.
 
I am running 14.2p2 on the latest repo on my laptop, a Thinkpad W520 with the i915 driver. I initially attempted to upgrade my laptop and desktop (nvidia driver, also 14.2p2 on the latest repo) from plasma5 to plasma6, about the time that pkg 2.0.0 came out. I was able to roll back to a previous boot environment, and have attempted the upgrade several times. I followed the UPGRADING document, (pkg remove -f plasma5-plasma && pkg install plasma6-plasma; pkg autoremove, followed by a pkg upgrade and reboot).

Each time, the upgrade completes, but I am unable to get KDE to start. I try both the X11 and the wayland versions, and neither will start. With the X11 instance, I get a black screen with a mouse cursor. With plasma-wayland, I get a blank screen and no mouse cursor...

Another thing I noticed is that from SDDM (which reconfigured itself from my settings from before the upgrade), I no longer have access to any of the virtual terminals. So VT1-VT8 do not seem to exist any longer.

So is it possible to get KDE upgraded, preferably without reinstalling my system? Can anyone give me a point in the right direction?

Thanks,
--vr
Upgrading using pkg-update(8) is not advisable.

The dependency hell (both Linux and BSDs, anything that runs KDE, suffer from it) makes it way too messy. There's just way too many details in place on a nice install of KDE 5. Even with KDE 5, if you want the next version (like 5.20 to 5.27), it really means a clean reinstall of the whole machine from scratch.

If you want to install 6 cleanly, that also means reinstalling the whole machine from scratch. No way around it. If you try using any of the automated utilities out there, please be aware that they only work on a very limited subset of the details. Resolving the inevitable incompatibilities between library versions is a never-ending task that recursively expands, that's what dependency hell is. To avoid dependency hell, you're much better off just doing a clean reinstall using the latest components from the ports.

BTW, when you use SDDM, it runs on one of the VTYs, true. Yeah, it's not very easy to access the other VTYs while SDDM is running, but those VTYs don't disappear. They are still there, still functional. If you want to access them and use them anyway - that is technically still possible, but figuring out the exact steps, and what to even look for - that is left as an exercise for the reader. The only hint I will provide is: Google is your friend.

So my advice: Just reinstall from scratch. And don't worry about Plasma Wayland for now, that stuff still needs to get a lot of ducks in a row before it becomes usable.
 
Upgrading using pkg-update(8) is not advisable.

The dependency hell (both Linux and BSDs, anything that runs KDE, suffer from it) makes it way too messy. There's just way too many details in place on a nice install of KDE 5. Even with KDE 5, if you want the next version (like 5.20 to 5.27), it really means a clean reinstall of the whole machine from scratch.

If you want to install 6 cleanly, that also means reinstalling the whole machine from scratch. No way around it. If you try using any of the automated utilities out there, please be aware that they only work on a very limited subset of the details. Resolving the inevitable incompatibilities between library versions is a never-ending task that recursively expands, that's what dependency hell is. To avoid dependency hell, you're much better off just doing a clean reinstall using the latest components from the ports.

BTW, when you use SDDM, it runs on one of the VTYs, true. Yeah, it's not very easy to access the other VTYs while SDDM is running, but those VTYs don't disappear. They are still there, still functional. If you want to access them and use them anyway - that is technically still possible, but figuring out the exact steps, and what to even look for - that is left as an exercise for the reader. The only hint I will provide is: Google is your friend.

So my advice: Just reinstall from scratch. And don't worry about Plasma Wayland for now, that stuff still needs to get a lot of ducks in a row before it becomes usable.
Upgrading using pkg-update(8) is not advisable.

The dependency hell (both Linux and BSDs, anything that runs KDE, suffer from it) makes it way too messy. There's just way too many details in place on a nice install of KDE 5. Even with KDE 5, if you want the next version (like 5.20 to 5.27), it really means a clean reinstall of the whole machine from scratch.

If you want to install 6 cleanly, that also means reinstalling the whole machine from scratch. No way around it. If you try using any of the automated utilities out there, please be aware that they only work on a very limited subset of the details. Resolving the inevitable incompatibilities between library versions is a never-ending task that recursively expands, that's what dependency hell is. To avoid dependency hell, you're much better off just doing a clean reinstall using the latest components from the ports.

BTW, when you use SDDM, it runs on one of the VTYs, true. Yeah, it's not very easy to access the other VTYs while SDDM is running, but those VTYs don't disappear. They are still there, still functional. If you want to access them and use them anyway - that is technically still possible, but figuring out the exact steps, and what to even look for - that is left as an exercise for the reader. The only hint I will provide is: Google is your friend.

So my advice: Just reinstall from scratch. And don't worry about Plasma Wayland for now, that stuff still needs to get a lot of ducks in a row before it becomes usable.
Wait, reinstall the whole machine from scratch or be stuck where I am? That seems a bit like a microsoft solution.
 
Wait, reinstall the whole machine from scratch or be stuck where I am? That seems a bit like a microsoft solution.
Yeah, reinstall completely or be stuck.

But no, that's not a Microsoft-style solution. I do know that in Windows, it's possible to upgrade specific software packages and not be faced with reinstalling the entire OS. Yeah, the good stuff in Windoze would cost a pretty penny, and would come with stupid restrictions on how to use/install it. That's been the case since Windows 98.

In Windows, if you upgrade (not a clean reinstall, but an in-place upgrade) from win7 to win11, Office 2016 would not give a rat's ass, and run fine. Your hardware might want updated drivers, but those would be a minor issue. There was a weird hardware driver issue in windows 8, but that was short-lived.

This 'Reinstall completely or be stuck' is something that both Linux and BSD camps struggle with. I personally found that with FreeBSD, that is easier to tolerate.
 
Upgrading using pkg-update(8) is not advisable.

The dependency hell (both Linux and BSDs, anything that runs KDE, suffer from it) makes it way too messy. There's just way too many details in place on a nice install of KDE 5. Even with KDE 5, if you want the next version (like 5.20 to 5.27), it really means a clean reinstall of the whole machine from scratch.

If you want to install 6 cleanly, that also means reinstalling the whole machine from scratch. No way around it. If you try using any of the automated utilities out there, please be aware that they only work on a very limited subset of the details. Resolving the inevitable incompatibilities between library versions is a never-ending task that recursively expands, that's what dependency hell is. To avoid dependency hell, you're much better off just doing a clean reinstall using the latest components from the ports.

BTW, when you use SDDM, it runs on one of the VTYs, true. Yeah, it's not very easy to access the other VTYs while SDDM is running, but those VTYs don't disappear. They are still there, still functional. If you want to access them and use them anyway - that is technically still possible, but figuring out the exact steps, and what to even look for - that is left as an exercise for the reader. The only hint I will provide is: Google is your friend.

So my advice: Just reinstall from scratch. And don't worry about Plasma Wayland for now, that stuff still needs to get a lot of ducks in a row before it becomes usable.
it might have been, but I didn't notice the difference now. the set of porting problems persists anyway. rather, you need to pay attention to kms-drm (I built it myself on the updated system.) and sddm is run from ttys:
ttyv8 "/usr/local/bin/sddm" xterm on secure
 
it might have been, but I didn't notice the difference now. the set of porting problems persists anyway. rather, you need to pay attention to kms-drm (I built it myself on the updated system.) and sddm is run from ttys:
Funny, I have SDDM working perfect, even though I have an unchanged /etc/ttys:
Code:
ttyv8   "/usr/local/bin/xdm -nodaemon"  xterm   off secure
 
Funny, I have SDDM working perfect, even though I have an unchanged /etc/ttys:
Code:
ttyv8   "/usr/local/bin/xdm -nodaemon"  xterm   off secure
I can't start plasma 6 wayland if sddm starts from rc.conf. I think the issue needs to be studied due to the fact that I use radeonkms.

in my case, plasma6 behaves very unstable. I tried both updating the existing system and installing it from the system without any packages. There was only one result.
 
I can't start plasma 6 wayland if sddm starts from rc.conf. I think the issue needs to be studied due to the fact that I use radeonkms.

in my case, plasma6 behaves very unstable. I tried both updating the existing system and installing it from the system without any packages. There was only one result.
Well, I did mention earlier in this thread to not worry about Plasma 6 Wayland (in post # 12).

The 'Working SDDM' comment of mine applies to a machine that is running Xorg. Under Xorg, SDDM is fine, and handles Plasma 6 fine. Like I said earlier, there's still a lot of ducks to get in a row before Plasma 6 becomes usable with Wayland on FreeBSD.

It is a lot of work to recompile stuff with correct flags and to make sure ENV variables are the correct ones and that the ENV variables persist across reboots.
 
Yes, I understand that plasma 6 wayland is an experimental game. I did not put it correctly about stability, rather, the standard keyboard shortcuts are not filtered.
 
I have finally had a chance to revisit this, and I fond a video (
View: https://www.youtube.com/watch?v=MYLAV-mLJtk
) that discusses how to upgrade from 5 -> 6, so apparently it is possible to upgrade without a reinstall...However, I have not gotten the right incantation to get it to work. I did the upgrade based loosely on the video noted above, as well, I removed all of the kf5 packages, which also removed several of the plasma6 packages (including plasma5-plasma). So I reinstalled plasma6.

So here is my conundrum. When I try and start kde through sddm, I type my password, it grays out the user, but never attempts takes me to even a blank X screen. My logs are giving me errors like:
Code:
Apr  7 12:18:43 defiant kernel: sonewconn: pcb 0xfffff80260d9e540 (0.0.0.0:10050 (proto 6)): Listen queue overflow: 193 already in queue awaiting acceptance (1 occurrences), euid 122, rgid 122, jail 0

Apr  7 12:21:51 defiant kernel: pid 3073 (ksplashqml), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)

Apr  7 12:21:53 defiant kernel: pid 3105 (kded6), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)

And top has a boatload of drkonqu messages:
Code:
17124 storm         2  89    0   364M   126M RUN      7   0:02  90.32% drkonqi
17207 storm         2  55    0   364M   126M wait     2   0:03  89.45% drkonqi
17227 storm         2  45    0   364M   126M wait     5   0:02  81.81% drkonqi
17167 storm         2  88    0   364M   126M RUN      3   0:03  65.71% drkonqi
17245 storm         2  45    0   364M   126M wait     4   0:03  57.01% drkonqi
17321 storm         2  89    0   364M   126M CPU14   14   0:03  56.36% drkonqi
17318 storm         2  50    0   364M   126M wait    11   0:03  50.55% drkonqi
17247 storm         2  37    0   364M   126M wait     6   0:03  49.99% drkonqi
17415 storm         4  61    0   356M   136M wait     3   0:03  47.63% kwin_x11
17556 storm         2  58    0   364M   126M wait    10   0:02  46.57% drkonqi
17315 storm         2  24    0   364M   126M wait    14   0:02  40.32% drkonqi

drkonqi also dumps a boatload of error messages like
Code:
Apr  7 12:17:42 defiant kernel: pid 10542 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)
Apr  7 12:17:42 defiant kernel: pid 10547 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)
Apr  7 12:17:42 defiant kernel: pid 10550 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)
Apr  7 12:17:42 defiant kernel: pid 10551 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)

Can anyone suggest where to look to get this working? Because I have a conundrum, because I can't continue to run plasma5 and keep my system updated, since the devs have taken that option off the table.

Thanks,
--vr
 
I have finally had a chance to revisit this, and I fond a video (
View: https://www.youtube.com/watch?v=MYLAV-mLJtk
) that discusses how to upgrade from 5 -> 6, so apparently it is possible to upgrade without a reinstall...However, I have not gotten the right incantation to get it to work. I did the upgrade based loosely on the video noted above, as well, I removed all of the kf5 packages, which also removed several of the plasma6 packages (including plasma5-plasma). So I reinstalled plasma6.

So here is my conundrum. When I try and start kde through sddm, I type my password, it grays out the user, but never attempts takes me to even a blank X screen. My logs are giving me errors like:
Code:
Apr  7 12:18:43 defiant kernel: sonewconn: pcb 0xfffff80260d9e540 (0.0.0.0:10050 (proto 6)): Listen queue overflow: 193 already in queue awaiting acceptance (1 occurrences), euid 122, rgid 122, jail 0

Apr  7 12:21:51 defiant kernel: pid 3073 (ksplashqml), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)

Apr  7 12:21:53 defiant kernel: pid 3105 (kded6), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)

And top has a boatload of drkonqu messages:
Code:
17124 storm         2  89    0   364M   126M RUN      7   0:02  90.32% drkonqi
17207 storm         2  55    0   364M   126M wait     2   0:03  89.45% drkonqi
17227 storm         2  45    0   364M   126M wait     5   0:02  81.81% drkonqi
17167 storm         2  88    0   364M   126M RUN      3   0:03  65.71% drkonqi
17245 storm         2  45    0   364M   126M wait     4   0:03  57.01% drkonqi
17321 storm         2  89    0   364M   126M CPU14   14   0:03  56.36% drkonqi
17318 storm         2  50    0   364M   126M wait    11   0:03  50.55% drkonqi
17247 storm         2  37    0   364M   126M wait     6   0:03  49.99% drkonqi
17415 storm         4  61    0   356M   136M wait     3   0:03  47.63% kwin_x11
17556 storm         2  58    0   364M   126M wait    10   0:02  46.57% drkonqi
17315 storm         2  24    0   364M   126M wait    14   0:02  40.32% drkonqi

drkonqi also dumps a boatload of error messages like
Code:
Apr  7 12:17:42 defiant kernel: pid 10542 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)
Apr  7 12:17:42 defiant kernel: pid 10547 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)
Apr  7 12:17:42 defiant kernel: pid 10550 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)
Apr  7 12:17:42 defiant kernel: pid 10551 (drkonqi), jid 0, uid 1001: exited on signal 6 (no core dump - bad address)

Can anyone suggest where to look to get this working? Because I have a conundrum, because I can't continue to run plasma5 and keep my system updated, since the devs have taken that option off the table.

Thanks,
--vr
Don't trust YT, ppl who post there tend to not think things through, and will end up forgetting about something important that does need to be mentioned, because it's THE detail that made a difference for them - usually something that was taken care of before Step 0 of the very process they're explaining.
 
Okay, I finally managed to get it installed on my desktop machine. I ended up creating a saved boot environment, doing a pkg remove --glob "plasma[56]-*" "kf[56]-*" "qt[56]-*" (removed 414 packages) and a pkg autoremove (an additional 382) and then pkg install plasma6-plasma sddm. Now I am going back and getting all of the stuff that was removed.

A little trying, but I learned a couple of things.
Thanks,
--vr
 
Why did you remove the "6" packages?
Read the post carefully, and put together the sequence of actions.

This is an unfortunate necessity that arises out of dependency hell. It's easier to remove both 5 and 6, and then use dependency resolution to bring back the only 6 by installing an updated meta-package.

If 6 were not cleanly removed in the initial step, the dependency hell would bite and not upgrade from 6.2.1 to 6.3.2 properly. Dependency hell is not a beast that is easily tamed.

I think that it is a good idea to question the moves and understand the logic/motives behind them.
 
Why did you remove the "6" packages?
I removed everything. Then reinstalled the "6" packages...As was said in Aliens, "Nuke it from orbit...It's the only way to be sure." And if you read further up in this thread, the suggestion was made to do a complete reinstall. So this was a complete reinstall as far as KDE was concerned.\\

Read the post carefully, and put together the sequence of actions.

This is an unfortunate necessity that arises out of dependency hell. It's easier to remove both 5 and 6, and then use dependency resolution to bring back the only 6 by installing an updated meta-package.

If 6 were not cleanly removed in the initial step, the dependency hell would bite and not upgrade from 6.2.1 to 6.3.2 properly. Dependency hell is not a beast that is easily tamed.

I think that it is a good idea to question the moves and understand the logic/motives behind them.

Yes, and my timing is rather unfortunate, since I appear to have caught it during the 6.3.3 -> 6.3.4 build cycle....At least that is what I think is happening, since things are missing on freshports as well. I was able to install on my desktop yesterday afternoon, but when I tried on my laptop last night, the metapackage, plasma6-plasma was not found. and when I copied the 6.3.3 pkg from my desktop, it couldn't find the plasma6-discover. So I am waiting for a bit until 6.3.4 is done
 
Back
Top