Linux guy on Windows 11 and... liking it?!

Framework Laptop 13
1

I’m a Linux guy and have built what is, IMO, a really great Arch Linux Frame.work system - and the w0rld is perfect, right??? Well - I cloned that install so I can always go back and I did what [some] Linux users might think is sac-religious - I installed Windows 11!

And I really like it… you see, all of my infrastructure runs on *nix - my server machine runs ProxMox, my NAS is a Debian host running FreeNAS, and most of my services are either ProxMox containers or ProxMox Linux VMs - amongst others…

So I got to thinking that… on my iMac I run MacOS, and I still use, view, work with and RUN all my *nix services/machines right from there with no issue… and Windows has all that ability; and the Frame.work is much more supported on Windows - and so far I’ve been REALLY enjoying the flow. YES, I get a bit pissy that I can’t use ALL my *nix commands easily - try using RSYNC across a Linux machine to Windows… first, you have to use some odd version of RSYNC - and even if you do C:\Users\myuser\some\folder\and\file.ext converts super oddly. Something like:
rsync username@192.x.x.x:/my/nix/folder cygwin/f/some/stupid/path/thats/not/real

To say the least, some things work way more easy say from Linux to MacOS than Linux to Windows - but what does work [everything] IS impressive. I have tried to get MiraCast working on Arch Linux [and Wayland] for ages… its just not there yet, however in Windows I just press WIN+P and in 20 seconds I have a 2nd screen on my 60" Samsung… or most other TVs in my house. Intel graphics has a nice application to tweak things, and I can run Grand Theft Auto 5 on my Frame.work without too much hassle… YES, its super different from Linux - and I’ve lived there for more than a decade… but I don’t know why *nix people fight ‘Windows’ SO much.

I see that THIS hardware works much better on Windows, so I’m giving it a whirl. Just found ‘oh-my-posh’ for a nifty Powershell prompt and the WSL/WSL2 is very impressive. [My aforementioned rsync issues probably work more easier using the WSL, but I haven’t figured out just where all the folders/files LIVE within the WSL sub-system just yet… but I’m sure its easy enough.]

Any other *nix users enjoy using Windows on their Frame.work? And, if anyone has any tips for a very non-Windows user to get things that *nix people might like or enjoy, drop me a line…

For now, and I can’t believe I’m saying so, I’m really liking Windows on the Frame.work - I still USE all my *nix THINGS - most of my work gets DONE in *nix, but why fight the funk? What do you think?

pAULIE42o
. . . . . . . . .
/s

2 Likes
2

Is there something that you are particularly thinking of when you say that the framework hardware is better supported on Windows? As far as I know, Fedora 35 onwards fully supports the Framework hardware out-of-the-box. With the kernel parameters nvme.noacpi=1 acpi_osi='!Windows 2020' the s2idle performance is also up-to-snuff.

1 Like
3
  1. Frictional scaling in conjunction with user experience.
  2. Hardware acceleration (e.g. Firefox, without the need to set / modify config parameters)
  3. Wifi…without kernel regression issue.
  4. Hibernation, OOTB.

Difference between able to do something, vs doing something well, consistently and with minimal surprises.

1 Like
4

I came to a similar finding / conclusion quite early on. Linux, when compared with Windows, on the Framework Laptop was (and still is) too much of a friction piece to get going for day to day use.

This is because I have very little patience for “getting things to work”. It’s a tool, to me, it’s not a hobby. I have zero patience to “get it going”. For people who sees this as a hobby, sure, there’s sense of accomplishment when you can get it working (that’s really is the primitive part of the brain). Higher functional planning would ask “why did you spend so much time on trying to do x here, when doing the same x ‘there’ can take less time…allowing you to do more in less time”".

5

I don’t experience much of the above. Fractional scaling I don’t bother with, I scale applications on a per application basis. Accel works natively in Firefox Nightly, so I expect accel to arrive properly in Firefox 103. Wifi hasn’t been an issue for me, I’m not sure if openSUSE patches the firmware, but so far it’s been fine. Hibernation has also worked just fine on openSUSE Tumbleweed.

6

I would say that’s not the average user use case.

Not the average user use case.

Maybe. That variation in distro patching speaks to poor standarisation of hardware support at any given point in time.

What steps did you take to enable it? Like this? Oooh…so much friction.

The average users don’t want to be system administrators…just want to use and go.

Linux communities tend to somehow turn a blind eye to even recognize the friction to “just open and use”.

Going to call it a day now.

7

Well, Firefox, Discord, and other programs provide user friendly scaling buttons. “Not the average user use case” seems to be very wrong.

And I said I predict it to come in Firefox 103.

Ironically, it worked out of the box for me! Do you like pulling questions from 2009 to try and disprove something tested in 2022?

8

That’s content scaling, not app scaling, right? Sure content scaling is available by the apps implementation. …But you said you scale the applications.

9

I’m sorry I don’t have the absolute perfect terminology. But I think I got the point across well enough.

10

Where are you getting 2009 from? That’s from 2021.

This is another blog post from 2022:
Enable Hibernation in openSUSE Tumbleweed (linux32bituefi.blogspot.com)

11

Ah, I misread. Thanks for catching my mistake. However, I don’t know what to tell you. Hibernate works just fine without additional setup, in fact, let me test right now and grab system logs to show you!

[18694.749802] ACPI: PM: Waking up from system sleep state S4
[18694.760521] ACPI: EC: interrupt unblocked
[18694.944142] ACPI: EC: event unblocked
[ommited]
[18695.515349] PM: hibernation: Basic memory bitmaps freed
[18695.515357] OOM killer enabled.
[18695.515358] Restarting tasks ...
[18695.515632] usb 3-9: USB disconnect, device number 2
[18695.515583] mei_pxp 0000:00:16.0-fbf6fcf1-96cf-4e2e-a6a6-1bab8cbe36b1: bound 0000:00:02.0 (ops i915_pxp_tee_component_ops [i915])
[18695.517775] done.
[18695.518976] Bluetooth: hci0: Firmware timestamp 2022.17 buildtype 1 build 43435
[18695.520922] Bluetooth: hci0: Found device firmware: intel/ibt-0041-0041.sfi
[18695.521037] Bluetooth: hci0: Boot Address: 0x100800
[18695.521039] Bluetooth: hci0: Firmware Version: 171-17.22
[18695.521041] Bluetooth: hci0: Firmware already loaded
[18695.532639] PM: hibernation: hibernation exit
13

That’s sleep / suspend, not hibrenation.

Again, your lack of “absolute perfect terminology” (as you phrased it) failed you?

14

This time, I am 100% sure you’re wrong. I am pretty sure I hit the hibernate button, and I am pretty sure PM said hibernation exit

15

Something I don’t understand probably. How did that complete in less than 1 second?

Looks like it’s enabled by default since around April 2022?

Current documentation:
https://opensuse.github.io/openSUSE-docs-revamped-temp/power/

So seems like your stars are aligned to have it working OOTB…

16

It didn’t. The initial filesystem loaded up the hibernated session, and as such it held the old time. Most operations also complete EXTREMELY fast during a system’s startup procedure, and properly exiting hibernate before getting to a user usable interface is going to be faster than timing to the usable user interface.

17

I see. My comprehension failed me this time.

18

I just want to say that Intel graphics performance is one area where it should almost certainly be better supported on LInux, perhaps even substantially so.

That being said, it’s my impression that this difference is more dramatic with older Intel graphics - my friend was getting literally (not figuratively) 1 to 2fps (not a typo) on Sandy Bridge simply running an eduke32 game mod yet, under Linux, was easily getting 40-50fps.

(it’s worth noting this was with OpenGL which is basically the worst-case scenario for Intel graphics on Windows; and no, he wasn’t using the Microsoft-supplied graphics driver).

Fun fact: Ivy Bridge and Haswell has vulkan support on Linux, but not on Windows.

That being said, I have to wonder if a much more GUI-based Linux user like myself would want to throw Windows 11 away in an instant - it’s my impression that it’s the “terminal warrior” type that has a higher appreciation for modern Windows than those of us that know how to use the terminal well enough but have major trouble even remembering basic commands but can remember exactly where we placed a file located 10 sub-folders deep (which I’ve found actually lends itself kind of well to Linux’s use of config files for low-level system OS configurations and the like).

For me, GUI-focus is a big reason I chose Linux Mint - I would strongly argue that Linux Mint’s GUI is more similar to traditional pre-Win8 Windows than modern post-Win8 Windows is to traditional Windows, and I was saying that years before Win11 was a thing!

(on a somewhat similar subject, I get a lesser impression that the “software dev” type is more of the type of person to like modern GNOME while the very hardware-focused “can’t code worth a darn” type like myself would instantly throw away GNOME; I’d also argue that there’s a definite overlap between software devs and terminal warriors that doesn’t tend to overlap quite as often with the “hardware first” type of person; fun fact, none other than Jim Keller is actually quite non-savvy at using a PC - Jim @ Anandtech: “I’m like the least computer savvy computer designer I think. I just had somebody help me log in to one of my online accounts.”)

19

I wonder if that’s the case of the app not leveraging its execution environment (e.g. less effort put on the Windows side of the port).

Intel, in general, only actively support their hardware when there’s money to be made. i.e. older hardware simply drops out of support from Intel. (any any commercial support / brands would do the same).

Linux…hell, it supports hardware for 20+ years if needed by anyone willing to write code…with or without a cheque.

i.e. Linux, from a certain perspective, is the land of extended life support / second wind for old hardware…to die.

Intel’s wording on Linux support / graphics drivers:
These drivers are provided and maintained by the Linux* distribution vendors and not by Intel

20

I don’t know either.

At the end of the day, there’s strengths and weaknesses to every platform. Nothing is perfect. It comes down to picking the right tool, to accomplish the task in the most cost effective way possible (time, money, brain cells, sanity).

e.g. Love my Zwilling Kramer 10" chef knife…doesn’t mean I need to chop down a tree with it.

Or, you simply don’t ask Thor to wiz you a portal…something more suited for Dr. Strange.

Similarly, the correct strategy, for any ‘correct’ / well informed / effective tool selection (physical or digital) is to make sure your toolbox has the tool, and your skills are ready to take on the task. I run Linux, Windows and macOS. Sticking to a platform to do something that’s taking 2 hr when it could have been done in 30 minutes on another readily available platform is just insanity, not loyalty.

Options, people. Options and choices. Use them.

1 Like
21

This applies more to the userland drivers, not the kernel drivers. In either case, if it’s a point for Linux, I’ll take it =)