FreeBSD Foundation Flounders on 15 with Rust, pkgbase, and KDE

astyle, I just tried that. I have a seldom used machine, and I changed it to Wayland, with labwc. I left it alone for 30 minutes and when I came back to it, I hit enter, and the window manager was up and running.
Labwc by default does not have a power manager daemon. KDE does. astyle when I tried KDE under Wayland I also had problems with the power management, in my case I found the keyboard unresponsive and I had to power cycle my system. Not to mention that I have not been able to get it working with any DM.
 
astyle, I just tried that. I have a seldom used machine, and I changed it to Wayland, with labwc. I left it alone for 30 minutes and when I came back to it, I hit enter, and the window manager was up and running.
I'm more interested at this time in seeing if power management is working under KDE Wayland... And info I'm seeing these days suggests that it's a hit and miss. Some ppl say it's not a problem for them (but those ppl also have stuff compiled, have the very latest versions, and a lot of other details lined up), and some say that they're having problems with power management, too.

I am seeing progress towards improvement in that area, though.
 
I'm more interested at this time in seeing if power management is working under KDE Wayland...
Something like turning off the display after 5 minutes of idle?

I don't think I'd want the DE handling anything like CPU frequency or PCI power states/hardware stuff.
 
Windows 95's look is about the minimal I can deal with:
That was a good UI. Hate Microsoft as much as you want but you can't deny that Windows 95's design was very well thought out and perfect. Today's "Modern" GUI's and styles are terrible. They are slow, bloated, and have poor function/user experience. All for the sake of wanting to change because some UI designer wants to be different and radical. Even worse is how the engines behind those GUI's are all HTML, CSS, and JavaScript.

Microsoft spent a ton of money and time researching the Windows 95 UI. Smart people would copy as much as possible and continue to reuse that and coast off Microsoft's effort/money.

Really would like to see a return to form.
 
Today's "Modern" GUI's and styles are terrible. They are slow, bloated, and have poor function/user experience. All for the sake of wanting to change because some UI designer wants to be different and radical.
Really would like to see a return to form.
The solution is to port toolkits to Xt (X intrinsics toolkits). Motif was compatible with so much, in part because it used Xt. However, Motif was difficult to program for. GDK which Gtk2 & Gtk 3 use, needs to be ported to Xt.

In the future, the client end of X will stay as long as C or very close C is around. The client end of X will always be needed for X or Wayland.

Xt needs to be ported to libxcb. It hasn't happened, bc it's seen as an uphill battle, that everything else will have to be ported to it. Because X client will always be useful, it's worth the vision to convert it all to xcb.

X server isn't going anywhere soon. It may be X12 or X13, but it will still be fundamentally X. And at least, the client end will be useful even longer, because it will work on Wayland or whatever comes next.

As for Wayland and X, I predict that the servers might hybridize.
 
The solution is to port toolkits to Xt (X intrinsics toolkits). Motif was compatible with so much, in part because it used Xt. However, Motif was difficult to program for. GDK which Gtk2 & Gtk 3 use, needs to be ported to Xt.

In the future, the client end of X will stay as long as C or very close C is around. The client end of X will always be needed for X or Wayland.

Xt needs to be ported to libxcb. It hasn't happened, bc it's seen as an uphill battle, that everything else will have to be ported to it. Because X client will always be useful, it's worth the vision to convert it all to xcb.

X isn't going anywhere soon. And at least the client end will always be useful, even on Wayland. It may be X12 or X13, but it will still be fundamentally X.

As for Wayland and X, I predict that the servers might hybridize.
Excuse the dumb ignorant question here, but why have things like X11 or Wayland in the first place? I see during FreeBSD boot and pre-X11 environment on UEFI platforms, a nice graphics capable low latency fast frame buffer. Couldn't libraries be built to make use of that instead?
 
Excuse the dumb ignorant question here, but why have things like X11 or Wayland in the first place? I see during FreeBSD boot and pre-X11 environment on UEFI platforms, a nice graphics capable low latency fast frame buffer. Couldn't libraries be built to make use of that instead?
Idk, but wouldn't that be considered a more basic display server? Whatever it is, X client will be made to work on any such display server.
 
Something like turning off the display after 5 minutes of idle?
Correct. My problem is frankly session restore, and having the DE respond adequately to those 5 minutes of idle... I don't want my session crashing after 5 minutes of idle. What if I have something in Konsole or Firefox? This is where Pipewire comes in.

I don't think I'd want the DE handling anything like CPU frequency or PCI power states/hardware stuff.
I don't expect the DE to control that stuff, but displaying the info in a nice layout - now that's nice to have.
 
DE/WM and screen blanking.
If you do not need "screen locking" use xset:
xset dpms 0 0 600

Powers off the screen if idle for 10 minutes. whack a key, wakes screen up. Equivalent to pulling the cable to the monitor.

Screen Lock is a different matter, most DEs have a "screensaver/locker" utility or for most WMs, old fashioned xscreensaver.

Isn't Pipewire an audio thing like ALSA/PulseAudio? If so are you wanting "I'm playing music on my system, I want it to keep playing even when I walk away to get lunch?"
 
Excuse the dumb ignorant question here, but why have things like X11 or Wayland in the first place? I see during FreeBSD boot and pre-X11 environment on UEFI platforms, a nice graphics capable low latency fast frame buffer. Couldn't libraries be built to make use of that instead?
Xorg/X11 are the libs that build on that basic graphics capability in the first place. In simple terms, the graphics that are provided by BIOS andUEFI are at the bottom of the graphics stack, then come the GPU drivers like amdgpu/nv/i915, and then Xorg on top of that. There's a reason it's called a graphics stack... Wayland is an alternative to the aging Xorg/X11.
 
  • Like
Reactions: mer
Excuse the dumb ignorant question here, but why have things like X11 or Wayland in the first place? I see during FreeBSD boot and pre-X11 environment on UEFI platforms, a nice graphics capable low latency fast frame buffer. Couldn't libraries be built to make use of that instead?
Its a good question. Underneath most Xorg (and Wayland compositor) drivers is libdrm.

You can draw directly with it. I ported a gameboy emulator to do just that including raw keyboard access. Before that we were using syscons(4) and vgl(3). But the later was broken when the newer console system was introduced and subsequently removed.

The reason why we tend not to is because IPC is poor (programs need to share the framebuffer). X11 does way more than is needed for a consumer playing games or browsing the web. But for enterprise / hetrogeneous environments it is useful.

I do have an early implementation of vgl(3) ontop of libdrm but I don't really have time to finish it off and tend to not use FreeBSD for client stuff currently.
 
That was a good UI. Hate Microsoft as much as you want but you can't deny that Windows 95's design was very well thought out and perfect. Today's "Modern" GUI's and styles are terrible. They are slow, bloated, and have poor function/user experience. All for the sake of wanting to change because some UI designer wants to be different and radical. Even worse is how the engines behind those GUI's are all HTML, CSS, and JavaScript.

Microsoft spent a ton of money and time researching the Windows 95 UI. Smart people would copy as much as possible and continue to reuse that and coast off Microsoft's effort/money.

Really would like to see a return to form.
I couldn't stand Win95 when came out! My first GUI was GEM, then classic MacOS (called just System 6/7 back then) and on PC's I was very happy with W4WG 3.11 with Central Point Desktop. I used to know almost all Win3 .ini's by heart, and passionately hated W95 Registry (still do). I switched to W95 only because of Macromedia Freehand, when they stop supporting Win3 (AFAIR v7). BTW, Win95 was barely usable and horribly unstable OS, Win98SE was the first version that worked well.

As usability goes, I was most productive and fastest using MacOS 8/9, very rarely touched mouse for any work in Finder. I loved OS X because of Unix/tcsh, but as productivity in GUI goes, IMHO it was a huge step back. Somewhere around first versions of OS X, I deployed at work Mandrake as netatalk/samba server, it had KDE, and that's what I went with when I started with FreeBSD (~ 2006). I'm still using KDE on X11, I have no plans to use Wayland, I'll wait a bit more for XLibre port to iron out problems and I'll switch to that.
 
astyle, that gave me a QCTM (Quietly Chuckling To Myself) moment. I met my wife, who had AOL, and we started living together, I started spending time online, and somehow it led to me becoming a sysadmin. Ironically, she's not at all technical, and once we started living together, I used it far more than she did. I think she had even bought AOL For Dummies. Ahh, I feel young again.
 
Correct. My problem is frankly session restore, and having the DE respond adequately to those 5 minutes of idle... I don't want my session crashing after 5 minutes of idle. What if I have something in Konsole or Firefox? This is where Pipewire comes in.
Not having session restore could be played-off as Plasma being so stable it isn't expected to crash :p (sounds questionable for enterprise wanting stability though)

But not being able to do something simple like blank a screen is wild; is Wayland that important? Technical aside, it could be argued Plasma on Wayland might be less energy-efficient, use more power, and have higher energy costs :p
 
Not having session restore could be played-off as Plasma being so stable it isn't expected to crash :p (sounds questionable for enterprise wanting stability though)

But not being able to do something simple like blank a screen is wild; is Wayland that important? Technical aside, it could be argued Plasma on Wayland might be less energy-efficient, use more power, and have higher energy costs :p
Come on... under Xorg, I can walk away for 5 minutes to grab lunch... same is possible under Windows, MacOS, you name it. So why not Wayland? I do want feature and capability parity here. I am sold on Wayland, I do see technical benefits. 'Energy efficiency' is irrelevant here, sorry. I should be able to set time limits for the screen to turn itself off when I walk away or go to sleep myself... and I expect Wayland to behave accordingly (just like Xorg does).
 
I'm only surprised that you never tried to automate that 'lolcat' thingy. IDK about zsh, but in bash this will work:

exec > >(while IFS= read -d '' -r line; do if [[ -n $line ]]; then echo; printf "%s\n" "$line" | lolcat; fi; done); PROMPT_COMMAND='printf "\0"'

🤪

I do not want to have everything printed in rainbow all the time :)


vmisev said:
Seriously now, my only contact with Fluxbox was way back ~'06 in the first FreeBSD based distro I ever touched - .ua made FrenzyBSD LiveCD. Fluxbox was very usable, nicely configured and more than enough for LiveCD based OS. BTW, FrenzyBSD is "guilty" that I fell in love for life with FreeBSD. Soon after Frenzy, when I decided to go with actual FreeBSD on bare metal, I went with KDE because that's what I knew from my previous Mandrake use, and since then it's only console (vt, of course) or KDE 🤷‍♂️

Last time I used Fluxbox was 2008 I believe ...
 
So, in regards to the new GUI installer.... if we could have our way and convince developers, (knowing and accepting) that the only constraint is there must be a graphical installer. What would be the best toolkit and/or language to use? Even something that runs directly in the console without X11?
 
So, in regards to the new GUI installer.... if we could have our way and convince developers, (knowing and accepting) that the only constraint is there must be a graphical installer. What would be the best toolkit and/or language to use? Even something that runs directly in the console without X11?
In the FreeBSD Foundation article "Graphical Installer for FreeBSD" they stated that new graphical installer will be based on gbsddialog(1) which is based on Gtk+
 
So, in regards to the new GUI installer.... if we could have our way and convince developers, (knowing and accepting) that the only constraint is there must be a graphical installer. What would be the best toolkit and/or language to use? Even something that runs directly in the console without X11?
My guess is, that you were hinting at this all along. There's also graphics of an image through using Lua for the FreeBSD logo on the start up terminal.
why have things like X11 or Wayland in the first place? I see during FreeBSD boot and pre-X11 environment on UEFI platforms, a nice graphics capable low latency fast frame buffer. Couldn't libraries be built to make use of that instead?
wouldn't that be considered a more basic display server? Whatever it is, X client will be made to work on any such display server.
Underneath most Xorg (and Wayland compositor) drivers is libdrm.

You can draw directly with it. I ported a gameboy emulator to do just that including raw keyboard access. Before that we were using syscons(4) and vgl(3). But the later was broken when the newer console system was introduced and subsequently removed.

The reason why we tend not to is because IPC is poor (programs need to share the framebuffer). X11 does way more than is needed for a consumer playing games or browsing the web. But for enterprise / hetrogeneous environments it is useful.

I do have an early implementation of vgl(3) ontop of libdrm but I don't really have time to finish it off and tend to not use FreeBSD for client stuff currently.
It would have to work for VT now.

I've actually been curious how X compares to DOS which a program is started on the command line, and a single graphical application popped up.

For minimal of what exists, Xt and maybe Xaw3dxft, or libxcb.

Maybe a curses fork with a graphical appearance like how Lua is used for the Beastie splash screen.

If anything, aside from the installer, I'd like to see the libxcb client library on a lighter more modern implementation of X server collaborated on by BSD's for the native X server and client.
 
What I'm curious about is what is gonna be used as backend for Gtk+? Linux has GtkFB, "an implementation of GDK (and therefore GTK+) that runs on the Linux framebuffer. It runs in a single process that doesn't need X. It should run most GTK+ programs without any changes to the source." /q, but what IDK is how it will be done on FreeBSD?
 
In the FreeBSD Foundation article "Graphical Installer for FreeBSD" they stated that new graphical installer will be based on gbsddialog(1) which is based on Gtk+
Well, that blog entry has no date, and the Github link (provided by the author) is dead... And FreeBSD is supposed to be a professional OS for professionals... 😩 This looks harsh, but somebody at the Foundation needs to get their act together.
 
Back
Top