Does Desktop have a future on BSD?

Beastie7

Aspiring Daemon

Reaction score: 443
Messages: 591

I honestly don't see a problem with FreeBSD shipping an openbox-like wayland compositor. App support notwithstanding; it could give us a base display system we can all use and make our own. Openbox doesn't really assume anything; its just a blank screen which is great. An extension API would be needed though; for panels, menu bars, etc.
 

Snurg

Daemon

Reaction score: 577
Messages: 1,349

I honestly don't see a problem with FreeBSD shipping a ... wayland compositor.
Wayland... This is an interesting and dangerous topic.

I don't have anything against wayland, as long as it remains optional, so users are left the choice whether they want it or not.
A few years ago there was a discussion on the freebsd mailing list after the announcement of the introduction of libinput into FreeBSD 12.2.

I warned to keep this optional, as libinput does not fully support, actually breaks a lot of functionality regarding mice wheels.
This issue has become a long-running problem on all Linux distros since their introduction of libinput, and also on FreeBSD (and also this forums) since.

The bad thing with libinput now is, that the libinput developers implemented it in a way that makes it permanent, no matter whether you manually configure xorg to use evdev.
At least on Linux, it is being placed between the hardware and the xorg event stuff, so the only way to remove it is to modify the sources.
This ensures maximum difficulties for users wanting to remove it.

Personally I guess there are strong pro-Wayland forces in the FreeBSD core team, as my request to include libinput in a way that it can easily be removed, for example by build flags, resulted me in getting moderated on the mailing list.
 

Beastie7

Aspiring Daemon

Reaction score: 443
Messages: 591

Wayland... This is an interesting and dangerous topic.

Major app providers are starting to provide backend support Wayland. We don't really have a choice if we want to remain relevant. I think the protocol is incomplete, but it does solve vital UX problems. When Nvidia (!), OBS, KDE, Mozilla, and the likes are starting to tip-toe into that world; that's a sign we should follow suit.
 

mickey

Aspiring Daemon

Reaction score: 296
Messages: 744

1. Did you research if that chipset is supported in stable or 13-beta?
2. It's open source. Unless that chipset needs special instrumentation & initialization, you can easily add that device ID to the list & build the module from source. Did you send a patch or the missing device ID to FreeBSD's bugzilla, to help to improve this unsatisfying situation?
1) No, it's for a friend's notebook who just wants to use that thing (without cable), and going for stable or 13-beta is just not a viable option.
2) As soon as I get positive feedback from her, that the modified kernel module works, I will create a PR for sure, or most likely.
 

Snurg

Daemon

Reaction score: 577
Messages: 1,349

They can be low-quality but are widely used and it is better to have support/driver in FreeBSD. The integrated NIC in many motherboards is RTL.
True... I was ranting because of my experiences (shame on me). What I am thinking of is how good it would be if there were more quality-awareness, for example that the users know that they should not expect skyrocket performance from Realtek devices.

Major app providers are starting to provide backend support Wayland. We don't really have a choice if we want to remain relevant. I think the protocol is incomplete, but it does solve vital UX problems. When Nvidia (!), OBS, KDE, Mozilla, and the likes are starting to tip-toe into that world; that's a sign we should follow suit.
True, too!
What I wish however, is that users who do not need it (yet) have sort of easy way to opt out, to avoid being affected by the "Wayland childhood issues".
 

kpedersen

Daemon

Reaction score: 1,641
Messages: 2,466

Major app providers are starting to provide backend support Wayland. We don't really have a choice if we want to remain relevant.
I am assuming that neither Windows or macOS is focusing on Wayland and yet both of these guys are going to remain relevant. So I am not convinced that relevance is related to supporting a Linux centric display system.
 

Beastie7

Aspiring Daemon

Reaction score: 443
Messages: 591

I am assuming that neither Windows or macOS is focusing on Wayland and yet both of these guys are going to remain relevant. So I am not convinced that relevance is related to supporting a Linux centric display system.

Open Source desktop upstream is Linux. Our relevance is tied to it. Unless, like I’ve stated before, we have our own full graphics/toolkit stack we have no other choice. As crappy as it sounds, they currently decide our destiny on the desktop.
 

Mjölnir

Daemon

Reaction score: 1,482
Messages: 2,114

I am assuming that neither Windows or macOS is focusing on Wayland and yet both of these guys are going to remain relevant. So I am not convinced that relevance is related to supporting a Linux centric display system.
IIUC Android is Linux. ChromeOS, too? I thought it is. Look @ their market share & more importantly: it's growth. Look at the trends, not at the status quo. Which display system does this new GiggleOS & FirefoxOS use? What do them Game-boxes use? These boxes are sold in high numbers. As are consumer tablets & laptops & smartTVs. My settop box (cable <-> TV & radio) runs Linux inside, sadly so.
 

Zirias

Daemon

Reaction score: 1,147
Messages: 2,073

Open Source desktop upstream is Linux.
Open Source desktop upstream is the projects actually building Open Source desktops (e.g. KDE) or even "just" window managers as well as display solutions (e.g. X.Org).

Unless, like I’ve stated before, we have our own full graphics/toolkit stack
X.Org is still alive and works on many systems (even including Windows). It might have its drawbacks, but there is NO pressing need to come up with something new. I don't think anyone would oppose something new if it is really better for important usecases AND written in a sane, portable way.
 

Beastie7

Aspiring Daemon

Reaction score: 443
Messages: 591

Open Source desktop upstream is the projects actually building Open Source desktops (e.g. KDE) or even "just" window managers as well as display solutions (e.g. X.Org).


X.Org is still alive and works on many systems (even including Windows). It might have its drawbacks, but there is NO pressing need to come up with something new. I don't think anyone would oppose something new if it is really better for important usecases AND written in a sane, portable way.

You guys aren't understanding what I'm saying here. Linux-centric maintainers, community boards, and developers target Linux first, not Linux and FreeBSD. Once they abandon X.org; who's going to maintain it? The FreeBSD committers? Us? I highly doubt that. Either we follow or we make our own solutions; thus our own ecosystem.
 

kpedersen

Daemon

Reaction score: 1,641
Messages: 2,466

You guys aren't understanding what I'm saying here. Linux-centric maintainers, community boards, and developers target Linux first, not Linux and FreeBSD. Once they abandon X.org; who's going to maintain it? The FreeBSD committers? Us? I highly doubt that. Either we follow or we make our own solutions; thus our own ecosystem.
I understand your concern. However I can just as likely see us sticking with X11 and waiting for Wayland to get replaced. Xorg has also demonstrated that it doesn't need much maintenance. After all, many Wayland fans have been stating that it hasn't been maintained for over 5 years. A few more (decades) won't hurt.

Especially now that drivers are now more tightly bound to the kernel rather than Xorg, we already have our modesetting driver.

We will replace X11 one day. I honestly don't think it will be Wayland however. I don't even really dislike Wayland. It is simple and stupid (in a good way). It just doesn't really solve enough. We will have an abstraction layer soon enough once Wayland matures.

IIUC Android is Linux. ChromeOS, too? I thought it is. Look @ their market share & more importantly: it's growth. Look at the trends, not at the status quo. Which display system does this new GiggleOS & FirefoxOS use? What do them Game-boxes use? These boxes are sold in high numbers. As are consumer tablets & laptops & smartTVs. My settop box (cable <-> TV & radio) runs Linux inside, sadly so.

I think this is quite a useful example. Wayland was made to improve set top boxes and the embedded world. Skipping some of the more complex network awareness and other layers. And then Android still doesn't use Wayland. In some ways I would see FreeBSD consider Android's SurfaceFlinger display system instead of Wayland. Especially if it went primarily with popularity. Both are open-source, SurfaceFlinger is far more mature and commonly used.
 

Beastie7

Aspiring Daemon

Reaction score: 443
Messages: 591

While some projects might think "Linux is the world", others still understand diversity and open standards, valuing portability. See for example https://freebsd.kde.org/

KDE is one of very few exceptions.

I understand your concern. However I can just as likely see us sticking with X11 and waiting for Wayland to get replaced. Xorg has also demonstrated that it doesn't need much maintenance. After all, many Wayland fans have been stating that it hasn't been maintained for over 5 years. A few more (decades) won't hurt.

Especially now that drivers are now more tightly bound to the kernel rather than Xorg, we already have our modesetting driver.

We will replace X11 one day. I honestly don't think it will be Wayland however. I don't even really dislike Wayland. It is simple and stupid (in a good way). It just doesn't really solve enough. We will have an abstraction layer soon enough once Wayland matures.

I guess we can always throw a new user into the sharks and tell them to follow vermadens desktop guide. I'm sure that'll keep us relevant.
 

Zirias

Daemon

Reaction score: 1,147
Messages: 2,073

I honestly don't think it will be Wayland however. I don't even really dislike Wayland. It is simple and stupid (in a good way). It just doesn't really solve enough.
This is btw a summary I wouldn't have expected in THIS thread. Just because it really makes sense. The basic idea of wayland makes a lot of sense, because X indeed has a myriad of features nobody wants/needs any more, while still lacking features that would be important (e.g. a solid and secure solution for "locking" a display session).

I don't know exactly what the pain points with Wayland on FreeBSD are. All I say is there is no pressing need to abandon X.Org right now.
 

kpedersen

Daemon

Reaction score: 1,641
Messages: 2,466

KDE is one of very few exceptions.
Out of the 3 main desktop environments people use, KDE represents a fair chunk. Plus Gnome was invented to be a GNU/Linux or GNU/Hurd desktop "GNU Network Object Model Environment" (originally). They don't represent the only free desktops around. Only desktops that have tied themselves down to Gtk really have an issue.

I guess we can always throw a new user into the sharks and tell them to follow vermadens desktop guide. I'm sure that'll keep us relevant.
So long as FreeBSD has developers (not necessarily users), it will remain relevant. Compared to your average Linux distro (Arch Linux has <30 developers), FreeBSD is doing much better.

It is a bit of a warped view I know, but as for the desktop, even Linux is irrelevant. That doesn't really have any impact on being able to make good use of it does it?
 

kpedersen

Daemon

Reaction score: 1,641
Messages: 2,466

I don't know exactly what the pain points with Wayland on FreeBSD are. All I say is there is no pressing need to abandon X.Org right now.
I think we already have it in ports and working. I don't think it was a particularly hard port either.

Just like on Linux, 98% of users stick to X11. The other 2% are Gnome 3 users who left it to default to Wayland ;)

Edit:
Yes, here it is.

With Wayland the real issue is the terrible incomplete compositors that need to do far too much than an individual can maintain themselves (X11 window managers are just about possible). However we do actually have sway in ports too (IMO Wayland is basically Gnome or Sway until it matures):
 

sidetone

Daemon

Reaction score: 713
Messages: 1,557

I wish they would keep Xorg, and only replace parts of it with the best components of Wayland. Then keep it portable and simple enough where simple traditional window managers, and XCB and Xlib applications can still be used, perhaps through a layer or parts of Xorg itself. I don't think Wayland should fully replace Xorg parts that allow traditional programs and window managers to work. For the vast majority of programs, including window managers, not being able to use them anymore because of a switch to Wayland doesn't make sense.

Parts of anything, including Wayland, that have parts of Gnome added to it, become messed up and stay that way. Influence by Gnome will mess up anything and make it into a perpetual dependency-hell pile of bloat. I wish they would keep that separate.

KDE and XFCE components aren't like that, but should also remain separate with their own separate implementations on Wayland.


When something like Xorg no longer needs updates, I only wish to verify periodically that it doesn't need updates and is still maintained or audited from vulnerabilities.


Here's a good article that shows how XCB (X-protocol C-language binding library) is a compatible improvement (or partial to full replacement) over Xlib: https://www.informit.com/articles/article.aspx?p=1395423. According to this as of 2009, the downside of XCB was lack of documentation and obsolete documentation.
 

sidetone

Daemon

Reaction score: 713
Messages: 1,557

Linux-centric maintainers, community boards, and developers target Linux first, not Linux and FreeBSD. Once they abandon X.org; who's going to maintain it? The FreeBSD committers? Us? I highly doubt that. Either we follow or we make our own solutions; thus our own ecosystem.
The best way to have an ecosystem, is to have a loose commonality ecosystem within different BSD's (FreeBSD, DragonflyBSD, OpenBSD, NetBSD). Spread out maintenance throughout similar operating systems that have a keep it simple ideology. Especially for hardware drivers and display servers.
 

Beastie7

Aspiring Daemon

Reaction score: 443
Messages: 591

It is a bit of a warped view I know, but as for the desktop, even Linux is irrelevant. That doesn't really have any impact on being able to make good use of it does it?

Linux is treated as the open source desktop. It doesn't bode well for alternatives. It's actually pretty damn stressful for someone who's trying to get rid of macOS entirely.

The best way to have an ecosystem, is to have a loose commonality ecosystem within different BSD's (FreeBSD, DragonflyBSD, OpenBSD, NetBSD). Spread out maintenance throughout similar operating systems that have a keep it simple ideology. Especially for hardware drivers and display servers.

Ideally, but a mere pipe dream. Mainly because Wayland is so damn incomplete you have to write a lot of OS-specific code for even basic display services. Like display enumeration, screen recording, locking, etc. NetBSD doesn't have support for libinput too I believe either.

I have to hand it to the OpenBSD folks for deciding to run their own stew of X.org; that is their parachute. That's not a bad idea either for FreeBSD.
 

sidetone

Daemon

Reaction score: 713
Messages: 1,557

I have to hand it to the OpenBSD folks for deciding to run their own stew of X.org; that is their parachute. That's not a bad idea either for FreeBSD.
Is OpenBSD's Xenocara too locked down to be of use as an Xorg implementation on FreeBSD? I know you weren't saying Xenocara specifically, but as in to use that as an example. When the old Xorg stops being used by the Linux community, that can simply be taken over.

Xorg is required to have standards that include obsolete hardware. Xenocara sticks by same standards as Xorg. The only thing I would like to see is a layer that replaces Xorg's utilities required for obsolete hardware, or simple generic drivers for old hardware.


Xorg is actually pretty good. Recently the session manager on my computer has been broken, and I don't know if that's from my own set up, or if it's from recent adjustments to Xorg and XDM itself.
 

vigole

Daemon

Reaction score: 1,183
Messages: 1,100

I have to hand it to the OpenBSD folks for deciding to run their own stew of X.org; that is their parachute. That's not a bad idea either for FreeBSD.
That's the main reason, once in while we hear that OpenBSD has better desktop support. Let's imagine you're a neophyte:
(Note: I'm not for desktop or against desktop. Also It's not my concern whether FreeBSD is suitable for random Joe or not)

You go to the OpenBSD site, download the biggest image, and when you're installing, you tick every single options. When
you reboot the system you will find yourself in a GUI environments. Now from a neophyte point of view you have desktop.

Same neophyte comes to the FreeBSD website, again download the biggest image, start to install, and check every boxes.
He reboots the system and will drop in CLI environment. Then he comes to the Forums, opens a threat, so forth and so on.

From neophyte perspective, i.e. download the largest ISO and tick everything, OpenBSD is better for desktop than FreeBSD.
That's the idea of desktop support from his perspective. It's not about driver and supporting hardware, nor supplying ports.
 

vigole

Daemon

Reaction score: 1,183
Messages: 1,100

Final note:
We can come here, talking about desktop all-day. But here's the most important point, which is missing from this thread, and every single similar threads.
If you want The FreeBSD - Desktop Edition, you have to pick the proper communication channel. i.e. talk to these fellas at FreeBSD Project and Foundation:

Core Team

Board of Directors

Staff
 

Beastie7

Aspiring Daemon

Reaction score: 443
Messages: 591

Final note:
We can come here, talking about desktop all-day. But here's the most important point, which is missing from this thread, and every single similar threads.
If you want The FreeBSD - Desktop Edition, you have to pick the proper communication channel. i.e. talk to these fellas at FreeBSD Project and Foundation:

Core Team

Board of Directors

Staff

I wonder if the DRM work is complete/stable. I read an article Ed Maste wanted to give WiFi drivers the same treatment. Hopefully people will actually do the work; it’s honestly embarrassing. Given this is a professional, research driven operating system.
 
Top