As someone who also has a Linux desktop and laptops in addition to the Windows machine, WSL is terrible next to the real thing. Unless you need software that only runs on Windows, I advise against using it as a reason to stay on Windows.
I am now reaching 36, I just like 'things that work' - like Windows. Windows works with all my hardware, with all my 7 laptops, with all my 2 towers, I can install Visual Studio 2022 on it, I can install Ubuntu in WSL2 and I can play Dota2/Civ/Baldur's Gate/Warhammer 40k Insquisitor Martyr/listen to Spotify/debugging/deploying to Azure/writing an email/writing a Presentation in Power Point.
And then go back to my family - because not worrying about Wayland, that new driver or that new update or "that new thing "- suprsingly leaves a lot of time that you can spend with your family :)
And of course I am using the .NET stack with Angular- all smooth, all good, no hassle, just smooth sailing :)
- a desktop on Arch
- an arm-based NAS on Debian
- a router running OpenWRT
- a backup target on Debian
- a media player (x86) with Kodi/Wine/Dosbox/Mame/other emulators
All are in regular use, but only the NAS and router run 24/7.
- One very light (and slow by today's standards) Fedora laptop I use mostly from the couch and used to go with me to conferences.
- One reasonably sized (also old, but still snappy) Ubuntu laptop I use tethered to a desk with a keyboard and monitor.
- One that's mostly a file server that all computers on the network can see.
- One that's a "write-only" file server for backups.
What about the times when windows take forever to "getting things ready". I once spent around two hours updating windows. And just after the reboot there were ten new updates.
https://changewindows.org/platforms/pc/releases/windows-11-c...
[Version 10.0.22538.1010]
I had ZERO (0) issues with upgrades or downtime.
I too could make coffee or read a book while apt or dnf do their magic, but the truth is I can continue working while they do it, probably because, unlike Windows, they can delete and replace open files while they install, making the update-on-reboot issue on Windows a non-issue on Linux machines.
It's been a while since I had a BSOD on Windows, but I have been forced to restart it quite a few times when something didn't work or stopped working, or the VPN went crazy or some other malfunction. Unscheduled downtime is still very much an issue on workstations and I wouldn't want to look into Windows Server and how it handles this issue.
They also break Intel ANS on every single release - i.e. VLANs and LACP will stop working until Intel releases a new version. For Windows 11, Intel gave up and there will be no ANS anymore[1].
And the last, january 2022 update? It casually killed L2TP...
[1] https://www.intel.com/content/www/us/en/support/articles/000...
My NixOS machine updates in a few minutes even when it needs to compile some custom things. And it will not change settings without me changing the configuration. If it does, I can just reboot to the older version and file a bug report...
Edit: has anybody tried to install Windows on ZFS or Btrfs? If update does something you don't like, just boot to a snapshot and get rid of the update.
I used to do banking on a VirtualBox immutable disk image. I only made it mutable when it was time to regular updates, then made it immutable again for using the bank and government software (in Brazil you had some govt issued tools that made life easier). You can also make it work with filesystem snapshots as well as letting VirtualBox manage its own snapshots for you.
I don't game all that much; most of what I play that doesn't have a Linux-native build runs fine through Proton on my laptop. If I did do any serious gaming, I'd probably reluctantly have a separate Windows gaming rig that I'd use solely for gaming, but otherwise avoid it like the plague. Life is too short to deal with user-hostile OSes.
I get that my use-case doesn't apply to everyone, but I want people to know that if you do mostly web stuff etc... even as a non technical person, Linux can "just work".
Then my father's printer and sound card stopped working after upgrading to Windows 8, and I moved him to Linux as well.
In all this time there's only one thing I couldn't fix for them: the library's audiobooks had some form of Adobe DRM on them, and of course it didn't support Linux. I told them to take up the issue with their library, but of course they didn't have a satisfactory resolution to the problem.
Ironically I switched away from Windows and onto Linux precisely because I was sick of things not working on Windows. Sure, it has the best desktop application support of any operating system but to use it for anything more than just a desktop machine requires hours of configuration because everything is tailored to supporting technologically illeterate people (eg file extensions being hidden). Very little of any of this config can be automated -- or at least not in any easy way. And then Microsoft will pull some bullshit move that undoes half your config and you need to start all over again (much like the author of this submission has experienced). Or every few years you need to upgrade to a new release of Windows and you have to find where to change all those settings again plus learn a stack of new settings you need to change too.
Eventually I realised I was spending as much time maintaining the OS as I was actually writing software on it. So I stopped using Windows. Or rather decided to keep Windows around for emergencies but use Linux as my primary desktop (instead of keeping Linux around for emergencies and using Windows as my primary desktop).
My experience with Linux has been that you have the same install running for 10+ years (this is especially true for Arch because you have fewer risky distribution upgrades), but you can keep your config in Git so it's sync'ed across multiple machines or kept safe if you do need to reinstall your OS. Everything is automated and just works. Plus these days more and more software is browser based so you don't run into nearly as many "Linux not supported" issues as you did when I first gave up on Windows (back when XP was new).
I'm not saying your experiences are wrong but just offering the counterpoint that "thing that work" might differ for different people.
Instead, you run into "Firefox not supported" issues.
Arch is needy; it commits seppuku if you abandon it for too long. Every update on Arch is a "risky distribution upgrade", and it insists that you update every day.
On the other hand, I have kept Debian installs going for a great many years, with dist-upgrades never failing, and I know that some people have maintained the same Debian install since they first installed it in the 90s.
> zcat /var/log/pacman.log.1.gz-2018010214.backup | head
[2009-02-23 18:14] installed filesystem (2009.01-1)
You can hedge against any distro breaking by using a volume manager which allows snapshots of the root volume, or even better, creates these snapshots automatically before a distro upgrade. Both LVM and Btrfs can do this. I use LVM because I love boring tech.
It's certainly introducing more risk but these days I only update my Arch systems a maximum of once every 6 months (usually less than that) and never have an issue. This meme largely came about from ~10 years ago with the filesystem changes and migration to systemd. But it's been a pretty smooth ride since.
> Every update on Arch is a "risky distribution upgrade",
In theory it is a possibility but in practice that's not really the case. I mean you're not going to get breaking changes every week ;) Arch's package manager has really improved in recent years too so a lot of occasions that might have needed manual intervention can be handled automatically by the package manager (which wasn't the case back when this meme was true).
Also there is a news section on Arch's website so if you're genuinely nervous about running `pacman` you can check there first (this is the recommended best practice but most people don't bother these days because it's rare to have an issue).
> and it insists that you update every day.
Only if you kick off your package manager everyday. But this is true for any Linux distro since they're all pushing updated packages everyday so I don't really see how Arch differs in that regard.
> On the other hand, I have kept Debian installs going for a great many years, with dist-upgrades never failing, and I know that some people have maintained the same Debian install since they first installed it in the 90s.
I've had a Debian dist-upgrade fail before but that's just once out of a dozen(?) of successful dist-upgrades so I do agree Debian is solid. Never had an issue with upgrading CentOS either. Nor FreeBSD.
The only thing Arch does better than the above is you don't have an hour(s) of downtime during the upgrade. But as you said, that's paired with the risk that you might run into a breaking change at any point in the future. For that reason alone I wouldn't run Arch on servers (or at least not professionally -- thrown Arch on some personal servers in the past)
It's not a meme. Just this week I attempted to update an Arch system that had been left unattended for only a couple years. It got caught in some kind of cycle where it didn't have the GPG keys required to download the latest GPG keys. Then there was some package conflict issue. Finally after much wailing and gnashing of teeth I completed the update and rebooted... "kernel not found".
I gave up on it.
And lest I be accused of incompetence, here's another Arch user's 3-year update story, only a few weeks ago, on our very own HN:
"Somehow the upgrade didn't completely hose everything. It just took a bit of finagling in the form of looking through any packages that gave errors (errors like cyclic dependency stuff and conflicts from bad use of pip) and uninstalling those that I didn't see a need for or switching them to the package manager instead of pip. About an hour of babysitting it and I had a completely updated and working Arch setup." -- https://news.ycombinator.com/item?id=29872053
So I'm sorry, but "pretty smooth ride" my ass.
Yeah, if you're not going to update Arch in 3 years then I'd expect some issues but in fairness to Arch, the duration you're talking about here is literally the entire support life for some distros. So I certainly wouldn't call Arch "needy" for that. I mean there does need to be a cut off point somewhere and the fact that it's still possible to recover from that state at all -- even if it isn't a smooth ride -- is pretty fscking impressive in my opinion.
So yeah, I agree that if you're the kind of person to install an OS and then not touch it again for three(!!!) years and expect it to still update then a bleeding edge rolling release distro isn't going to be your best choice. But nobody was suggesting that Arch (or any distro for that matter) is perfectly suited for every edge case out there. And while I'm sorry you've been burnt in the past, I do think the way you've presented your arguments here is rather disingenuous.
Funny how the best thing to have happened to Windows in the last 10-ish years is gaining the ability to install Linux on top ;)
Also making Python and Angular first class citizens in their dev stacks. Both of which are prominent open-source projects.
also, as someone who has to keep a windows machine for availability of certain applications that have no linux versions yet, the statement that "windows just works" is bs in my experience
> also, as someone who has to keep a windows machine for availability of certain applications that have no linux versions yet, the statement that "windows just works" is bs in my experience
For a very limited set of values of "works". ;-)
Eh, pretty sure it is mostly Office, specifically Outlook, that keep Windows alive. Glorious PC Gaming Master Race is relatively small, although they do spend a lot of money.
I use Windows for audio recording, mixing, mastering and gaming, here are my examples of Windows not "just work"ing:
My Windows bluescreens immediately after booting on my Threadripper build with an MSI MEG x399 if I don't have the wifi driver installed. On a new install I have to deactivate it in UEFI settings, then boot/install windows, then install the driver before reactivating it.
Windows Update tries to install certain useless crapware (nahimic) that I can't reject or properly uninstall without a few obscure hacks to trick it.
I use an external audio interface, if Windows goes to standby I have to unplug it and plug it back in, otherwise it won't detect it anymore (this does not happen on Fedora 35).
Also windows will occasionally just switch my default input device after reboots or going to standby. This behaviour is not consistent, so I have to check everytime before I use it for calls. This problem does not occur on Fedora 35.
I currently have 13 .NET redistributables and Windows runtimes installed that all required manual downloading and installing for various games and software to work.
It truly is, the only part on music production that is keeping me is not even the DAWs themselves (I mainly use Cubase and Reaper, and Reaper already offers a Linux build, which works well), it's the VST plugins, which rarely if ever offer Linux versions.
Other music software I make heavy use of: Transcribe!, which offers a linux build and Musescore, which is open source and runs on Linux as well.
looking for some beta users, to get feedback as we're thinking a cross-platform component on similar lines.
1: https://aspsecuritykit.net/guides/aspnet-policy-authorizatio...
Today I use Slackware 14.2 on a netbook and -current (almost 15 nowadays) as my desktop OS. No driver issues, every emulated game works. I use either text mode via framebuffer (I can even watch videos and run SDL2 based cool stuff like mednafen) or fluxbox.
I don't care on modern gaming, Slashem and IF killed modern graphical-intensive games but with a childish and boring script. Kinda like the current decade trend of Marvel superhero movies. I had more fun reading classical sci-fi comics from the 50's. BG was cool... at 18. Today I have no time to follow a 20000 pages long script based not on sequential reading, but on choices. Slashem has better and more inmersive RPG elements (Nethack was the king of emergent gameplay) and I can play it anywhere on any machine from a 486 or a Classic Mac. Mednafen does magic for my gf, two player 8-16 bits classics with a gamepad while I use the keyboard. It works even without X. No DLC's, no always online. Oh, and Jap nerds translated loads of games into English. I am not an English native, but for sure I can read English better than Kanji.
On media, I don't care on Spotify, I have curseradio-improved, yt-dlp, Jamendo and Radio Paradise, and MPV plays HD content from Youtube and may other sites perfectly. My offline music collection (streamripper to Radio Paradise) does it better with the "shuffle" function on mocp. Fire and forget.
Time for my family? No social media, no "always connected" bullshit. I don't have to follow 2000 notifications on the framebuffer as in a GUI or a smarthone. https://68k.news, https://lite.cnn.io, https://efe.com for my country's news and Slashdot/HN/Lobste.rs has anything I want to read. Even in text mode. I also have some good times with Gopher and magical.fish/sdf.org and the tildeverse. Condensed geekeery thanks to few members, code sharing and a low volume of posts. Heaven. And, maybe, bbs@uncensored.citadel.org via SSH from time to time. Ah, a lot of people posted personal blogs on Gopher, so I can read posts back to ~2007 with the Gopher revival, and I can read it on a slow pace. The protocol is fast enough and there's no lag to load the sites, OFC.
Visual Studio? Heh. Nowadays almost everything I do it's cloud containered bullshit so there's very little to code, I almost do (jim)tcl and a bit of C for fun.
At home, everything else it's scripted over with a script called "do_the_internet.sh" pulling and fetching rss, mail, usenet news and podcasts. Once a week. No info overload.
Ironically thanks to that I spend much less time with the computer than you thanks to scripting ;), so I can focus on different stuff such as reading long books and biking.
And, yes, I geek sometimes, OFC, but with a fast as hell Fluxbox netbook, I can chat everywhere thanks to SSH (mosh) and Bitlbee (IRC gateway to Telegram/Jabber and whatever Pidgin supports), and hack and slash with Slashem on top of a mountain by using the phone's wifi sharing.
No ads. No unwanted updates. No pop-ups. No crapware. Batteries last hours.
If you are on Windows, get a Gopher client and head to gopher://magical.fish and gopher://sdf.org. Your brain will thank you.
It boils down to which OS you want to be the "second-class citizen" on your machine. With Wine, it's Windows. With WSL, it's Linux.
That equivalence is only valid for WSL1. Since WSL2 runs the real Linux kernel on a VM, the equivalent for it on Linux would be running Windows on a VM.
Not quite. When Windows is on a VM, you mostly run only Windows-specific software on it. Within the VM, everything works mostly flawlessly. Most platform-neutral software will be running on the Linux side, which makes accessing data from the Linux environment a non-issue.
And don't even get me started on Windows-1252 and CRLF line endings.
WSL2 still has some issues with VPN software and my company doesn't support it on my corporate Windows laptop.
I am genuinely interested on knowing your reasons for saying WSL is terrible. Just for clarification, I'm a Linux/MacOS user but the times I've had to use WSL in the past whether experimenting at home or in some companies where I was force to use Windows, I though it is actually one of the most decent things Microsoft has created on its entire existence. It is easy to enable, to operate, it integrates well with the network stack of Windows (kubectl port-forwarding or docker run -p8000:80 works out of the box) it performs very decently (with exceptions on IO filesystem operations) etc.
It's just that it's poorly integrated with the host environment (because there is a host environment). On Linux and Macs, the Unix side is the machine itself. With WSL1, you have a Windows kernel pretending to be a Linux one (I kind of prefer this approach) but my home folder is not my Windows home folder, but terminal still opens the Windows home anyway. On WSL2 (which I can't use on my corporate laptop because the VPN software breaks something in DNS and networking I haven't had the time to debug) the Windows side needs a network share to get access to the WSL environment (which is not present until the environment starts - and there are other side effects whether WSL starts before the VPN or after).
A Linux box (or Mac) is a so much simpler (to understand what happens) Unix environment it's not even funny.
Having said that, when you overcome those impedance mismatches, it mostly works for the typical use cases. Like you mentioned, Docker works (I need to set a DOCKER_HOST variable to point to the Windows box). And Windows Terminal is excellent (double height and width chars are still missing, but they have overline support).
It's death by a thousand paper cuts.
I finally gave up on it and went back to using my underpowered OSX machine instead of continuing to try to get WSL2 to do the job.
Was it possible? Probably. But I'm not going to spend the time to figure it out. There's probably a correct way to install everything we needed, but I didn't find it, and I found numerous wrong ways.