Re: [Sawfish] Your future plans? (Wayland)

[John Lumby <johnlumby@xxxxxxxxxxx>]

I'm a bit late to this discussion and may have missed some of the suggestions,
but my understanding is that there is a more-or-less-almost-workable generic solution for running any existing Xorg-based window manager without having a full Xorg server running,   *and* without re-writing the window-manager source code to do compositing,   which is to use Xwayland in rootful mode to emulate Xorg server.
Here is the article which describes how to do it for fvwm3 and implies similar can be done for any other window manager


https://www.reddit.com/r/unixporn/comments/srb6km/labwcfvwm3_futureproofing_your_window_manager_a/?share_id=Utn8ywo7G3ojfILNVN1cz&utm_name=androidcss

I tried to set this up on an old system of mine and the experiment ended in total failure for a different reason :
https://www.linuxquestions.org/questions/slackware-14/xorg-development-effort-slowing-in-favour-of-wayland-4175731955/page16.html#post6490219

but there are other articles elsewhere all implying it should work.

On the pro side of the equation,     this is completely independent of what distribution you happen to be running  (although you will probably have to acquire and install extra software).      I don't run any kind of distribution-provided Desktop Environment such as Gnome or KDE,         and I would guess most or all sawfish users don't either  -   why would you do so if sawfish manages windows?

On the con side,    and it is probably a big one for sawfish users  --  sawfish customizations and xxx.jl lisp functions may not work as expected.    and even if they do,    the look and feel will be rather different.      How different is what I wanted to find out but currently do not have a computer on which I can experiment.

Has any sawfish user tried this and if so with what success?

Cheers,     John Lumby

_________________________________________________________________


From: Trevor Cordes <trevor@xxxxxxxxxxxxx>
Sent: March 29, 2024 3:50 AM
To: Robert 'Bobby' Zenz <Robert.Zenz@xxxxxxxxxxxxxx>
Cc: sawfish@xxxxxxxxxxxxxxxxxxx <sawfish@xxxxxxxxxxxxxxxxxxx>
Subject: Re: [Sawfish] Your future plans? (Wayland)
 
On 2024-03-28 Robert 'Bobby' Zenz wrote:
>
> You got some terminology mixed up there. GNOME has two "session
> options", X11 and Wayland. They **greatly** favor Wayland because
> "pixel perfect frames" or something. GTK can either draw to X11 or
> Wayland. What happens when you have GNOME-Wayland running and open a
> program which still uses X11 is that it is attached as Xwayland
> client. Basically a X11 server alone for that client. So it works,
> however, Xwayland is quiet limited in what it can do and not. For
> example, it can't sport window managers as the other windows are not
> exposed to programs.
>
> However, nothing stops you from using a different desktop environment,
> even a "cobbled together" one on top of a "bare" X11 session. GNOME is
> not the center of the Linux Desktop Universe, even if they try hard.

Thank you guys for your input.  I don't reddit, and no place seems as
apt as the sawfish mailing list to talk about sawfish.  So here we are,
circle-jerk and all!

Yes, I understood the no-x-for-gnome-session vs no-x difference.  But,
and I should have mentioned this: what got me antsy in the first place
is before the F41 gnome-x-session removal, F40 (which will be released
soon) is doing kde-x-session removal!  Reading about them ditching x
for kde, and then gnome in rapid succession gets my radar dinging.
It seems obvious that RH/Fedora are making huge pushes to just ditch X
completely.  And just like with systemd, RH sets the tone for the
entire Linux world, and Fedora is the advance guard.

(And no, I am not running sawfish with GNOME DE, I've been an XFCE guy
ever since GNOME went off the rails with UI design.)

I am glad to hear from you kind folk, though, that others concur the
wording means we can still enjoy X and XFCE/sawfish login sessions for a
while to come, at least through F41.  But the news articles about these
things are, perhaps purposely, vague as to our fate as the lowly "other"
DEs.

As a dev and longtime RH user I can see where it's all headed.  Within
short order they will deprecate any sort of X session at login.  Make
you use -compat pacakges.  Then 1 or 2 releases after that they'll just
remove X support completely from GTK or Qt or a major app.  Ya, I'm
speculating, but I've seen it all before, so I'd bet money I'm right.

The only question is speed/timing.  I think they want to do it in 4-8
Fedora releases (so 2-4 years).  That's something we need to start
planning for.  If it really is 10-15 years out, then, yes, I can
happily put my head back in the sand.  I really hope you're more right
than I.

> I always liked how the implementation of Arcan worked, because that
> sounded much more sane.

I'll look into it, but my first impression of that odd website is: "uh
what does this mean to me?".  What I really wanted to hear was that
some has-sawfish-mentality programmers are already working on a highly-
configurable/programmable compositor/wm for Wayland.  Doesn't have to be
sawfish or sawfish people, just people who want more than the
LCD Windows-like nonsense everyone else seems happy with.

It's a small sample size so far, but if the 2 responders so far don't
have an answer on the tip of their tongue, there may not be an answer
(one of my fears)...

> That last part is, ufff. Wayland is a moving target, and it is a
> fragmented target, too (compositor specific features, dumb
> client library decisions, ...). So "being done" like with Sawfish here
> is at least a decade off, maybe longer. That's not encouraging, I
> guess.

It will go faster than you think if the big players with an agenda push
it hard.  It's already pretty clear in every article's main talking
points that it's not so much logic as emotion being pushed here.  Every
single promoter of Wayland, and thus basically every article writer,
has to include the obligatory "x is insecure", "x is old", "x is slow".
Yet we never hear anyone define "insecure", and the rare articles to
address it focus on the keystroke issue.  Which to me is a non-issue
because if you can't trust your distro's other packages to not do evil
stuff, then grabbing your keystrokes is the least of your problems.  (We
are dealing with linux desktops here, not Android phones.)  As for old,
I prefer old.  And slow, ya on my Ryzen 9 I really care about that.
"Slow" hasn't been an issue for 15 years.

In fact, it's the same tack taken when systemd was trying to takeover.
It's disingenuous, but it shapes perception and thus becomes truth to a
critical mass of people.  Heck, even systemd had more pushback than
Wayland seems to be getting!  Maybe _they_ learned from their mistakes.

> Also, I don't like to think of myself as a helpless victim, because
> that's not true. I can always walk away.

Ya, of course, but a walk away from sawfish, or Linux entirely, comes
with great pain, and as you so aptly referred to, this really equals
time.  And I have better things to do with my time, same as the sawfish
devs.  For now.  In the end it may come down to what requires the least
amount of pain/time.  I have a philosophy: no change for change's sake.
Or the Jerry-McGuire philosophy: Show me the features (rationale).  Not
a single thing Wayland offers makes my life better or saves me time.

> be to simply keep using X11. Also remember that there is the BSD
> community, they didn't seem so keen on Wayland, either.

Yes, good point: BSD may provide a redoubt against Wayland.  I haven't
researched it yet, but it would seem to me that Wayland may not even be
possible on BSD unless they decided to change a bunch of other things
to be more Linux-like.  And last I checked that was pretty taboo with
that bunch.

And the distro-holdouts (the Devuans of Wayland) + BSD may offer
lower-time-investment solutions when Fedora does pull the X plug.
Whilst I currently have a one-box-does-most approach to my
server/workstation, I could foresee a 2-box approach or a Wayland-less
GUI host + Fedora-on-vm-for-everything-else solution.  And that, as you
pointed out, should be viable for a decade?!

--
Sawfish ML


--
Sawfish ML