Hacker News Reader

> The problems that GNOME 3.X has always had continue to be present -- chiefly the UI scale is too large on low-res screens (the machine I was testing with has a 1366x768 display)

It's not just Gnome that does that. There's an entire generation of designers that's cargo-culting mobile designs at the moment. Breeze, KDE's default theme, is equally space-wasting. They're great if you have a touch screen but it would be really cool if you could turn them off on systems without touch screens, i.e. about 99% of the current Linux installations I'm guessing :-).

If you're willing to fiddle around, there are a few good themes that don't waste so much space.

I totally agree with your overall point, but:

> about 99% of the current Linux installations

Not if you include Android!

In these contexts people usually mean "GNU/Linux desktop" when they say "Linux". It is not as if we have any meaningful choice of desktop environment on Android-land anyway.

> Not if you include Android!

How many Android installations running Gnome have you seen in the wild :-)?

The person you're replying to said scaling not spacing.

Very different issues. They may have optimal spacing but their scaling is being handled incorrectly by the system, or they may have perfect scaling but suboptimal spacing. Or they're both good or they both suck.

It's best not to conflate the two.

I could also criticize the spacing, but much ink has already been spilled on Gnome 3.X on both sides of the argument; I'm not looking to start a flame war.

However, I was under the impression that previous versions of Gnome supported non-integer scaling factors, which seems to be gone from current GTK versions(?). Just being able to set the scale to 0.5 or 0.75 would go a long way for people with lower res screens.

Even though I don't use Gnome myself, this still affects me, since many applications use GTK, and thus have massive Fisher Price buttons compared to other programs I use.

There are themes that address the spacing / padding problem. I want to try Nextwaita [1] when I'll find the time to upgrade my Ubuntu 16.04 system. I have a try to Ubuntu 18.04 on a separate partition and padding is really unbearable.

[1] https://www.gnome-look.org/p/1289376/

Furthermore the Nvidia driver and Gnome Shell work only at 40 Hz. The screen flickers noticeably at 60 Hz (Quadro K1000m). No problems on 16.04 with Gnome Flashback, which is GTK3 and Compiz. I didn't check 19.04 yet.

Themes are not supported: https://stopthemingmy.app/ and cannot be changed via the happy path GUI, though I did see that Debian 10 includes Gnome Tweak Tool out of the box.

Maybe theming is still supported in Ubuntu?

There is going to be a way as long it's based on CSS files on the file system. I definitely don't want the default theme, I prefer scrollbars and other things my way.

Developers of Linux apps should accept they have difficult users. I'll buy a Mac when I'll settle for UIs I don't like.

I believe Ubuntu also ships a tool for changing the theme in the default install. However the upstream (Gnome) is adversarial to this (https://news.ycombinator.com/item?id=20376442).

> If you're willing to fiddle around, there are a few good themes that don't waste so much space.

The point of this "first impressions" exercise was to explore the happy path.

> It's not just Gnome that does that. There's an entire generation of designers that's cargo-culting mobile designs at the moment. Breeze, KDE's default theme, is equally space-wasting.

I would agree with that sentiment, though I didn't want to get into that flame war today. I think a lot of other people have already written extensively on this subject and I don't feel I can contribute anything new. Here are some relevant articles...

Several of these are about CSDs, which I think are a big point of contention with respect to the design philosophy of GTK/Gnome 3.X. This list does not include much about the KDE/Qt side, which I'm less familiar with.

I would certainly like to expand the side of this list on the "Pro Gnome" side in the interest of a more fair and balanced debate.

Pro Gnome 3.x:

Client-Side Decorations Initiative - https://wiki.gnome.org/Initiatives/CSD

Restyling apps at scale - https://blogs.gnome.org/tbernard/2018/10/15/restyling-apps-a...

Introducing the CSD Initiative - https://blogs.gnome.org/tbernard/2018/01/26/csd-initiative/

Against Gnome 3.x:

Why do we keep building rotten foundations? - https://davmac.wordpress.com/2016/07/05/why-do-we-keep-build...

GNOME (et al): Rotting In Threes - https://igurublog.wordpress.com/2012/11/05/gnome-et-al-rotti...

Torvalds on Gnome - https://web.archive.org/web/20190313090648/https://plus.goog...

Sway and client side decorations - https://drewdevault.com/2018/01/27/Sway-and-client-side-deco...

Server side decorations and Wayland - https://blog.martin-graesslin.com/blog/2018/01/server-side-d...

The problem with GNOME 3 - https://felipec.wordpress.com/2013/06/12/the-problem-with-gn...

IIRC, the Debian installer will make your user account sudo if you leave the root password field blank during the installation wizard. If you provide a root password then your user is not made into a sudoer by default.

I believe that is the case. I probably was not paying close attention at that stage.

It's certainly not a difficult problem to fix, I was simply commenting on the fact that it had to be fixed, which has not been the case on most other distros I have used.

Unfortunately the bold decision to take on Wayland (due to Gnome) in the upstream debian buster isn't going to propagate to low end hardware like Raspberry Pi incl. Pi4.

Though the Buster based raspbian will be available, they will continue using the LXDE version as Gnome is too heavy for the Pi.

I had tried Wayland for Pi3 earlier in my effort to get a smoother desktop experience on the Raspberry Pi[1], it was the smoothest desktop environment among other options but unusable due to XWayland apps crashing due to framebuffer issues.

I've settled for HW accelerated, Xfce4, Arch Linux (btrfs/zswap) on RPi3.

[1]: https://abishekmuthian.com/getting-smoother-desktop-experien...

> * My user account wasn't added to the sudo group by default, and using `su` broke something with PATH (/sbin wasn't in PATH after running `su`).

The default profile for users (/etc/profile) hasn't included sbin paths for the entire time that I've used Debian, which always seemed a little weird to me, since there are definitely binaries in there that normal users can still use (e.g. lsmod).

I am aware that the normal user account does not include /sbin, but `su` not adding it to PATH is different from previous releases to my knowledge.

I am unsure of why this is the default behavior, however.

Simply `su` should not change the PATH; you have to start a login shell, as in `su -`.

Huh. I guess I just never noticed that... normally I do `sudo su`, which I believe does update PATH.

sudo tip: `sudo -i`!

Yeah, sudo su always annoys me: sudo -i if you want a login shell and sudo -s if you don’t (e.g. you cd-ed somewhere and forgot to become root, but don’t want to cd again because the path is long or you have done some complicated setup.

Sudo su -c 'command1|command2' makes sense if both commands need root but you want to stay normal user afterwards.

Then sudo bash -c '...'

"sudo su" is "launch su with superuser privileges", which is eh... weird to say the least. You probably wanted "sudo -i".

Yes, this is a change from Debian 9, due to Debian switching which implementation of 'su' it uses, moving to the one used by most other distributions. NEWS entry with details: https://sources.debian.org/src/util-linux/2.33.1-0.1/debian/...

Working link, since sources.debian.org seems to be having problems at the moment:

https://salsa.debian.org/debian/util-linux/blob/debian/2.33....

Use `su -` to get the path you want.

> Personally, I won't be using Wayland on my main machine once I update that as I'm a CWM user, and Wayland support probably isn't every going to happen for that.

I've been thinking about this myself, as I use a handful of programs which I am unwilling to give up but which Wayland will probably never properly support (intentionally; I'm fond of recording and playing back arbitrary input events, which Wayland has helpfully blocked for security reasons). My reading of the docs is that it's possible to run Wayland+xwayland as, effectively, a complete X server... well, it's not even a drop-in replacement, it's just an X server that happens to be implemented differently. So I think that if/when Xorg starts to become problematic to continue using, I can just swap it out for xwayland running in root window mode, and then run my X window manager and programs on top. Might need to mess with environment variables to mask Wayland so nothing tries to use it directly.

Anybody else have thoughts/plans on the future of X if Wayland really does take over?

My understanding is that every Wayland compositor automatically runs XWayland by default for compatibility reasons, and will do for the foreseeable future.

Xwayland is Xorg. It is one of the DDX backends available for Xorg, along with kdrive, XQuartz, Xephyr, the xfree86 backend, etc. If Xorg "becomes problematic to continue using", it will simply be removed. Xwayland will be removed too, since "nobody actually uses the legacy X protocol anymore". That's the goal, to keep X around for as long as is needed for "modern" toolkits and "modern" applications to transition fully to Wayland, before removing X entirely.

> Anybody else have thoughts/plans on the future of X if Wayland really does take over?

I have been thinking about this for a few years at least (happy i3wm user).

I am looking at moving to Sway https://github.com/swaywm/sway (drop in replacement) at some point as it seems the development is really healthy. sircmpwn (Drew DeVault) has done a lot of really nice contribution to the FOSS community, in particular wlroots https://github.com/swaywm/wlroots I also thoroughly enjoy reading his blog https://drewdevault.com (articles from there often get posted here) there has been many a time I've read something he's written and thought "hear hear" in my mind.

It is also impacting my purchasing choices in regard to going for AMDGPU instead of NVIDIA for my next system that I plan to buy in a few months (mostly waiting for 3rd party Navi cards at this point).

I wonder, why use Wayland at all? What are the benefits?

Because drawbacks are huge (large number of apps would need to be customized and tiling managers would probably need to rewritten completely)

I don't use it and I don't have any issues with my current Linux desktop (using xmonad).

> I don't use it and I don't have any issues with my current Linux desktop (using xmonad).

It looks like there's an effort to port it to Wayland https://github.com/waymonad/waymonad

The fact is X11 is being supported mostly by Fedora and they're going to stop using it too.

It is already in maintenance mode with no new releases scheduled. Eventually it will be completely unmaintained, probably around 2029 when RHEL8 support is dropped.

https://blogs.gnome.org/uraeus/2019/06/24/on-the-road-to-fed...

> I wonder, why use Wayland at all? What are the benefits?

Muh security and muh pixel perfect frames.

But mainly, Red Hat is going all in on Wayland and has decided that X11 is deprecated. So if you want to develop applications for Linux that will be compatible going forward you will target Wayland and not X. (Realistically, the toolkits will use Wayland and not X and you will target those. Eventually GTK and Qt will drop X support altogether.)

What do you mean by "pixel perfect frames"?

I watch movies in VLC and I don't complain about pixels :)

(BTW. I don't care much about security of my desktop behind 2 routers and a NAT - yeah I know I'm still a target, but there are easier ones).

> What do you mean by "pixel perfect frames"?

Wayland enthusiasts usually use "pixel-perfect frames" to mean lack of tearing -- what happens when the framebuffer is changed in the middle of the "electron beam's" scan down the display and you get half of one image and half of another. Eliminating tearing generally means forcing vsync, which means your framerate is capped at 60fps, and a few milliseconds of input latency are introduced -- meaning there's a bit of a tradeoff between the fastest desktop and the smoothest. (This is true regardless of display stack, whether Windows, Quartz, X or Wayland). Xorg drivers can often run in a "tear-free" mode, but Wayland blah blah modern architecture blah blah designed from the ground up blah blah. Plus, Nobody Ever Uses the X protocol anymore. (Basically, you don't have a choice -- Wayland compositors are required to force vsync and operate in tear-free mode.)

Security is, largely, a non-issue unless you run untrusted binaries all the time and/or open ports on your X server to random places that can then run random programs on your display. However, for the paranoid (and a little healthy paranoia is a good thing), it would be more involved but still not a whole lot of work[0] to modify the X server architecture to support a more Wayland-like security model by default, while still allowing trusted apps to do things like take screenshots and respond to global hotkeys.

The advantages of Wayland thus boil down to "it's what the community wants". And by "the community" I largely mean Red Hat and the GNOME Foundation. Basically, younger programmers who were weaned on Macs or Windows want a more Mac- or Windows-like experience than X, which offers a more Unix-workstation experience, provides. So they decided that the graphics stack needs to optimize for local applications running on a composited desktop with fancy effects -- not local and remote clients all drawing to one display. Plus it gives them the opportunity to do what every programmer longs to do -- a Great Rewrite From Scratch, using Modern Frameworks and Modern Techniques because the end result will be so much better and this never turns out badly.

But you're going to have to deal with Wayland regardless of whether it offers advantages to you, personally, again, because our benevolent lords at Red Hat have spoken on the matter and they say X is dead.

[0] https://keithp.com/x-ideas/x-security/

> I wonder, why use Wayland at all? What are the benefits?

The question is "why write Wayland at all", and the answer is that re-inventing wheels is more fun than maintaining them and This Time We'll Get It Right™.

And then, after another decade or so, when enough corner cases have been dealt with that Wayland finally offers the usability of X circa 2009, a new bunch of kids will look at it and say "what's all the cruft in this codebase? Why all these messy corner cases? Let's start from scratch!"

And the Circle of Free Software will begin again.

> * My user account wasn't added to the sudo group by default, and using `su` broke something with PATH (/sbin wasn't in PATH after running `su`).

It's been like this since day one; people started to expect a different behavior after Ubuntu became popular, but Debian is consistent in this regard. Adding an account to the sudo/wheel group should be a conscious decision on the part of the administrator as it has major security implications.

Care to file bugs? We can possibly fix these in a point release..

I feel like a lot of these are a little vague to write proper bug reports. If I get around to it and can re-produce the behavior that I was seeing with software center, that would probably be narrow enough in scope to file a bug report. But other issues like "the search screen is slow" are probably rather broad.

They're really not. Bugs would be entirely appropriate.

I hate "plus ones" but... +1 to this.

Ok. I'm not very familiar with the Debian community. Is the place to put these types of bug reports the mailing list? Is there a Bugzilla?

BTS:

https://www.debian.org/Bugs/

How to report a bug: https://www.debian.org/Bugs/Reporting

There is an interface package, "reportbug", to assist in filing:

https://packages.debian.org/stretch/reportbug

And also debbugs for querying / working with the BTS:

https://packages.debian.org/jessie/debbugs

The GNOME network configuration interface really kills it for me. Ubuntu's old default was nm-connection-editor and I liked it a lot more.

I would cite the OpenBSD approach[1] as being particularly good. Ultimately, what you want is a list associating SSIDs with passwords, and that's all it is. It does not handle WPA2 enterprise though, which is sometimes a pain.

I'm using nm-connection-editor and nmtui (depending on m mood) on my daily driver, and will probably continue to do so for the foreseeable future.

1 - https://man.openbsd.org/hostname.if