You all donated en masse to have me use Windows 11 for a month, and so I did. What was it like for a long-time Linux user to go back and experience Windows as it exists now? Is it really as bad as we’ve collectively made it out to be? Did my month with Windows 11 consist of nothing but pain and misery, or are there good things to say, too? Or, was it an unexpected pleasant surprise? And ultimately, did I stay with Windows 11, or move back to the Linux world?
➡️ Donate through Ko-Fi ➡️ Donate through SEPA transfer* ➡️ Buy merch from our store ➡️ Why a fundraiser?
*Name: Thom Holwerda – IBAN: SE08 8000 0820 1684 4657 8414 – BIC: SWEDSESS
This year, I’m celebrating the milestone of having posted 20000 stories on OSNews during my 21 years as managing editor of OSNews. This is my full-time job, and since nobody is going to give me any bonuses, stock options, or golden pens, we’re running a big fundraiser to keep OSNews going. To add some spice to the whole thing, I added some incentives, with the first being using Windows 11 for a month. We’re slowly but steadily approaching the next incentive, too, which is a proper video tour of my office, (unique) computers, and massive devices collection. There’s a similar incentive to this Windows 11 one, but for macOS. Yikes.
The rules for the Windows 11 incentive are simple: use stock Windows 11 for a month for my computing tasks (with the exception of gaming – converting my Linux gaming PC to Windows just to play the same games seemed silly). I wasn’t allowed to use any debloating tools, but as an EU citizen, I do have the ability to remove a ton of Windows stuff thanks to the success of the Digital Markets Act. I also tried to stick to Microsoft’s own applications as much as possible, for that true “ecosystem experience”, and wasn’t allowed to hack my way into a normal local user account. I was all-in.
So what was it like?
Setting it all up
The installation process posed a number of challenges and issues. First and foremost, the Windows 11 installation process is incredibly barebones, and basically assumes no other operating system exists in the world. It has no clue anything other than Windows’ filesystems exist, making it dangerously easy to accidentally damage or outright delete any other operating systems you might have installed. My laptop happens to have two M.2 SSDs in, so I could safely dedicate one of them to Windows 11 without interfering with the other SSD with Fedora installed on it, but if you’re experimenting with Windows 11 on your Linux machine with just one drive, you might want to reconsider.
I also had to perform the first portion of the installation process – the WinPE section – with just my keyboard, since apparently, my trackpad was not supported and did not work at all. Once the system went through its first of what would be many reboots to come and loaded into the phase of the installation where you’re actually already running Windows 11, my trackpad came to life, but without any gestures support – so no scrolling. Not a gamebreaker or anything, but definitely annoying.
A bigger issue was that the Wi-Fi 7 Intel BE200 chip in my laptop was not supported out of the box by Windows 11. This meant that I had to install these drivers during the installation process, which involves going to the Intel website and finding the correct drivers to use. To make this process more obtuse and less intuitive, you can’t use the normal driver installer; you have to specifically opt for the “Intel® PROSet/Wireless Software and Wi-Fi Drivers for IT Administrators“, download the ZIP, unpack it on a different computer, put the unpacked drivers on a USB stick, and point the Windows 11 installer to this USB stick.
Mind you, the BE200 chip was launched almost three years ago, and there’s no excuse for Windows 11 not supporting this chip out of the box – like Linux does.
The remainder of the installation process involved dodging a lot of tracking and telemetry prompts, reboots, a lot of waiting, setting up the dreaded online account, waiting some more, and then finally ending up at the desktop. I then set out to enjoy my EU privileges by removing whatever applications I didn’t need and turning off features I didn’t want, as well as making sure all the drivers were up to date. This mostly involved installing the Intel Driver & Support Assistant and the Intel graphics drivers. Curiously, this is where I hit a returning issue: after installing the Intel GPU drivers for the first time, as well as after every subsequent update, the screen would go black and stay that way, forcing a reboot. Windows’ graphics stack is supposed to be able to gracefully handle driver updates, but clearly, some bug or problem was preventing the updated Intel driver from being reinitialised.
Once those initial setup tasks were behind me, I experienced two more problems. First, sleep/wake was entirely broken and simply did not work. It turns out Windows 11 really doesn’t like S3 sleep, and I had to specifically go into my laptop’s Dasharo Coreboot firmware to switch to S0ix get sleep/wake to work on Windows 11. Windows defaults to something it calls “Modern Standby”, which requires the S0ix state to be enabled. You can also disable Modern Standby which would presumably make sleep/wake work with S3 (?), but this is a whole ordeal and clearly not something Microsoft wants you to do.
Of course, the correct way of handling this would be for Windows 11 to adapt its sleep/wake settings to what the firmware reports, but alas.
Another problem were the laptop’s cooling fans seemingly leading lives of their own, spinning up loudly at entirely random times, irrespective of use. It was so bad and loud I assumed the laptop was damaged somehow, and nothing I tried alleviated the issue. However, a day after installation, a massive Windows update came in that somehow fixed the issue, taming the fans back to the normal levels that I had come to expect while running Linux.
Except for one curious problem that seems to tie the fan and sleep/wake problems together: roughly one out of three sleep cycles, Windows would spin up the fans to maximum blast, for long periods of time before actually going to sleep; on some occasions, sleep would never set in at all, forcing a reboot as the screen wouldn’t come back on either. This seems to be a widely reported problem on a whole slew of different hardware configurations, so I’m assuming Windows 11 is just trash at putting devices to sleep properly.
Note that this same laptop running Fedora Linux has none of these issues; sleep/wake works perfectly every time regardless of whether Coreboot is set to S3 or S0ix, and the fans behave exactly as you’d expect.
One thing I found almost too hard to believe was that Windows 11 apparently does not natively support the “US (int’l with AltGr dead keys)” keyboard layout. Instead, the only option it seems to have for the “US (int’l)” keyboard layout family is the one with regular dead keys, which I personally find unusable. For those that don’t know, dead keys are when you press e.g. ', but nothing happens until you press a letter which then gets the diacritic added to it: ' followed by e will turn into é.
You might spot the problem here: you often need to use characters like ' and " as actual characters, especially when you type a lot of English, but if they function as dead keys you have to hit them twice to use them as individual characters instead. This is incredibly annoying – way more than it seems on paper – so an alternative exists: “US (int’l with AltGr dead keys)”. On this keyboard layout, AltGr acts a modifier you need to press to turn certain keys into dead keys. To input é using this layout, you hit AltGr + ' followed by e.
This keyboard layout has been available as an option in every Linux installer and every desktop environment for as long as I can remember, so I never even considered it might not be available in Windows. Luckily, people have created third-party “US (int’l with AltGr dead keys)” layouts for Windows, so I ended up downloading this one, which works perfectly.
Input crisis averted.
I also ran into a few smaller issues. Windows’ window manager is incredibly limiting and dumb, and won’t even allow you to change things like titlebar actions. By default, double-clicking a titlebar will maximise a window, but I’m a BeOS user at heart and double-click titlebars to minimise windows (I never maximise a window). I kept accidentally maximising windows when I was trying to minimise them, which wasn’t pleasant. The fact that such basic settings virtually every operating system and desktop environment support are unavailable on Windows is indefensible.
Another pain point is Explorer, Windows’ file manager. It takes longer to load than a file manager should, and lacks basic features like dealing with compressed files – I don’t count a decades-old cumbersome wizard-style interface with countless steps to go through just to unpack a compressed file to be even remotely acceptable in 2026. Dolphin and Nautilus handle compressed files entirely transparently and much faster than Explorer does, and once you’re used to that, going back to ’90s style compressed file management almost feels insulting.
A quick non-exhaustive rundown of even more issues: Windows operating system updates are slow, cumbersome, and require way too many reboots. The Start menu desperately needs to be more customisable and adaptable to user needs. The widgets system in the taskbar is useless. The overview/Exposé feature drops frames all the time. I was never given an option to change my home folder’s name. There are way too many useless default folders in your home directory, and most of them you can’t delete (they keep automatically reappearing). Dark mode is still broken, with many dialogs and panels only available in light mode.
I also happened to run into a curious bug in Explorer where the icons in the Quick Access tab were fuzzy. No amount of troubleshooting could fix this. I admit this bothered me way more than it should.
Applications
As part of the incentive, I also wanted to experience proper Windows applications. First and foremost, this means using Microsoft Edge. Like many other browsers today – even Firefox – Edge spams you with useless “AI” nonsense you have to meticulously disable, but once you’ve done that song and dance, Edge is mostly just fine? I even felt like it did a better job of handing online video – less heat, less fan noise – than Firefox did, but I didn’t do any benchmarking or anything so I have no data to back it up.
The email situation on Windows is abysmal. You’re supposed to use the “new” Outlook, which is basically just a web application that also happens to send all your login credentials, emails, and personal information to Microsoft as a requirement before you can use it. While the irony of Gmail users complaining about this isn’t lost on me – email is not, never has been, and never will be a private medium – it’s still just unethical, unpleasant, and wholly unnecessary. To make matters worse, if you don’t have some sort of Office 365 subscription, Outlook even shows you ads. The new Outlook is just a long string of own goals before kickoff.
Nevertheless, I took my assignment seriously, and after choosing to ignore it’s just a website, after sending all my data to Microsoft, and after paying the cheapest possible Office 365 subscription offer I could find to get rid of the ads, I found that the new Outlook is, much like Edge, fine. While I’m sure it falls apart quickly for people with more advanced email needs, it handled my basic personal send-and-receive use case just fine.
If you disregard it’s a website that sends all your emails and personal information to Microsoft and that you have to pay for it even after paying for Windows itself, then yes, it is mostly fine. A ringing endorsement if there ever was one, isn’t it? This whole situation is criminal, and the clearest example of just how much Microsoft utterly despises Windows and its users. A desktop operating system needs to come with a solid, serviceable email client. I consider this non-optional.
Moving beyond Microsoft’s own applications, the application ecosystem on Windows is in a dire state. Anything developed over the last decade or so using the long list of modern frameworks and APIs Microsoft championed and subsequently abandoned is an exercise in frustration; most applications in this category are unfinished, buggy, slow and/or abandoned. Applications with more pedigree from the classic Win32 days feel outdated and out of place, but at least they tend to get the job done. The end result is an incredibly inconsistent, messy, and jarring user experience where every application clearly feels of its time, dependent on which set of frameworks and UI design philosophies Microsoft was pushing at that particular moment in time.
No two titlebars are of the same height. There are countless entirely different designs for titlebar buttons. The modern desktop context menu has its own classic Win32 context menu. Win32 applications look and behave differently than WinUI 3 applications which look and behave differently than Fluent applications which look and behave differently than Metro applications which look and behave differently than – and so on. No two applications have their important UI elements in the same place, and no two applications seem to be using the same design language. Hell, Win32 UIs use completely different-looking font rendering than “modern” UIs. The word “mess” doesn’t even begin to describe it.
As someone who is used to KDE and GNOME, whose developers still take consistency in both look and behaviour quite seriously, this is the single biggest reason why using Windows 11 was such a frustrating experience for me. It’s like reading a book where every few words, the language and script randomly change. I know UI consistency has been a dirty word ever since the web and then iOS rose to prominence – I lamented the death of consistency in UI design back 2012, which is fourteen years ago! – but the situation on Windows today is particularly dire.
Managing applications is also not as nice and effortless as it is on Linux. Most of the time, you have to manually browse around and download applications (and hope they’re not malware), which use one of an endless variety of different installation wizards, and then update these manually using countless different update services running in the background. There’s also a Windows Store, but its selection is limited. On top of all that, Windows also has its own very limited and basic package manager now, but it doesn’t come with an easy-to-use graphical user interface; you have to find and download one yourself, and it seems UniGetUI is one the more popular ones. It’s a mess of an application – with its own entirely unique titlebar and buttons, as is Windows tradition – but at least it works.
Keeping track of all the individual updaters, the Windows Store, WinGet, and so on is a massive chore, and a huge regression compared to what’s been the norm in the Linux world for a very long time. Desktop Linux solved keeping applications updated decades ago. Microsoft seems to be making it worse every time they add another different application delivery and management framework.
Windows applications are also absolutely obsessed with the system tray. It seems like every single thing you install wants to bury itself in the system tray, even when they’re not actually running. Before you know it, you’ll have a long string of random icons in there competing for your attention, and each seems to operate and behave a little differently than the other. Some open their main window when you click on them once, some when you click on them twice, some open a menu, some only respond by opening a menu when you left-click on them instead.
Of course, the menus that pop up all have different designs, as is tradition.
It’s not all bad, I guess?
There were positive aspects to Windows 11, too. It’s taken them a very long time, but with most of the various settings and configuration panels now moved from the old Control Panel to the Settings application, I think the latter has come into its own quite nicely. If you ignore the various ads for Microsoft’s services – a common tactic in commercial operating systems like macOS, Windows, and iOS these days – I find it quite easy to use. There’s always going to be some arbitrariness to the organisation and hierarchy of the various settings and panels, but overall, I found things relatively easy to find, and performance didn’t seem to be an issue.
Windows 11 also has a combined emoji/symbol picker now (Super + .), negating the need to dive into the Character Map, a horrid application which basically hasn’t been meaningfully updated since Windows 3.x. There’s an actual clipboard manager in Windows too now (Super + v), and it works great as well. These are two relatively recent additions that make some of the menial tasks related to text input quite a bit more pleasant.
I really don’t have much more to add to this measly “positive vibes only” section. Like Linux, Windows 11 found and set up our crappy HP Wi-Fi printer/scanner combo thing without any issues, I guess?
Did I stay with Windows 11?
No. Of course not.
I gave it an honest-to-god try. I put in the time, work, and even some money. I was strict, didn’t allow myself to do any non-gaming tasks on Linux, and truly used Windows 11 exclusively for a month. Whenever I experienced a short stretch of time where I felt “perhaps this isn’t so bad?”, one (or multiple) of the problems and issues described above would snap me out of it. For someone used to desktop Linux, where respect for the user, consistency, customisability, and performance are still held in high regard, Windows 11 feels like an endless string of punches in the face.
Whether I use a KDE or GNOME desktop, things look, feel, and behave consistently. There are no ads for services I don’t want, no online accounts forced down my throat, no dark patterns to trick me into subscriptions I don’t want. Managing and updating applications and the operating system are so effortless you barely even notice it’s happening, and whether I’m using an older machine or something brand new, performance is going to be good, and consistent. Desktop Linux is also going to respect my privacy, and I don’t have to worry about data harvesting.
Windows 11 just cannot compete with any of that, and my month with Windows 11 proved that to me beyond a shadow of a doubt.
Bravo, Thom, you got through it! After you paid for Office 365, did you try any of the applications? Despite not having had any meaningful changes in over a decade, Office in Windows also basically works if you avoid using the AI slop. I’d have liked your thoughts.
My day job uses Office and Windows 11, so it is a welcome thing to come home to Linux and KDE, which I’ve configured to look as much as I can like Windows 2000. I imagine you felt the same thing after your 30 days. Some of your grievances can be lessened with tweaks and utilities, like Microsoft’s own PowerToys. If you’re up to it, I’d like your thoughts on those someday.
Microsoft products today are basically an addiction for the American business world, and thus the U.S. government, as you know well. And Microsoft knows it, and acts like the abusive partner.
What is your next challenge?
“The tattoo ! The tattoo ! The tattoo !”
A great write-up. Thank you for your scouting and reporting.
I work with a small volunteer team supporting older folks (like me) trying to work with their computers – everything is a Windows or MacOS or iOS machine. I’ve been using Windows since Win95/3.x/NT/etc. and Linux for about 10 years – now completely on Linux (Mint).
I’ve recently refused to help with MS or Apple products since they cause confusion and frustration when the UI changes – which it does way too frequently.
Wtf is the Dasharo Coreboot firmware? Pretty sure that’s responsible for all your troubles.
In general, I like you comments and style, but honestly: this one is rubbish.
Haha, that’s fair.
But, I didn’t look into UEFI replacements myself, and do suspect it is to blame for a lot of the setup woes.
But I am bothered by UEFI being a back box, especially with the creepy Ring -3 stuff. Is Dashero actually good?
Well played and kudos: a mature reaction to a friendly banter, cheers mate!
I actually don’t know Dashero, but Linux apparently worked while Windows stuttered — and they forced that UEFI garbage on us. That alone makes them guilty.
what they forced on us is having to use microsoft certificates in the root chain of trust. not that i see many workable alternatives for that issue, at least that scale beyond a few tech savvy users.
uefi in itself is progress imo, i wasted to much time waiting for add-in cards with bootroms to initualize, the press a magic key combo to get in that card’s config menu, etc….
inphobia,
The secure boot cert situation sucks. The secure boot standard itself provided no way for owners to control it, which is an objectively stupid oversight, But the problem is that microsoft developed secure boot for itself. The needs of power users and alt-os were just never represented. We deserve a better standard, alas this is the standard we got.
It was only after the fact that backlash pressured microsoft to give in and let owners have a switch (on x86 anyway). Because user control was never (and still isn’t) part of the standard it remains very difficult for alt-os owners to use secure boot for themselves without reverting to microsoft keys.
Centralization of power is evil. At least microsoft are signing linux bootloaders so they’ll work on windows certified hardware, but I still find the fact that they have this much power over it to be rather appalling.
alfman,
seems i can’t reply directly to your reply (while i could on another post of yours), guess there’s a limit to how deeply replies can be nested on the site.
anyway, while i’m not salty about linux getting a ms signed bootloader i do feel this only enforces the microsoft/apple/linux oligopoly. for us “alt alt os” users running openbsd or plan9 (or perhaps gnu’s herd) it might actually be a tiny step backwards.
inphobia,
Yeah this was a regression when osnews migrated to wordpress, but seeing as you registered after that you probably never noticed 🙂
I see what you mean about being a problem for other operating systems, however not all linux operating systems work either since many are not signed by MS. My own distro isn’t and I did have the misfortune of encountering a laptop that refused to boot my own linux distro. I did my due diligence and read the reviews for it, but most if not all reviewers stating that “linux works” are posted by people running a signed mainstream linux distro and didn’t try to run an unsigned one.
Fortunately it’s rare to encounter x86 hardware that deprives owners of the ability to turn off secure boot.
Very nice article, Thom. Pleasant to read and easily to align with.
I have to work on Windows sometimes on behalf of customers and every time I end up with neck injuries from shaking my head.
One thing I also noticed: Linux people are far more open to the fact, that “our” applications are just one possible implementation of standards and protocols. Window users believe, that there is only their own way of doing things and anything else was not possible or wrong.
I spent 3 hours in a “Teams” meeting with a client, when suddenly their shitty Fortinet SSL VPN did not allow me to connect anymore. I pointed out from experience, that its not about the password but that my account had expired/got disabled.
They insisted in “Sharing Screen” (what would that help, when you just get access denied on client side and that’s it) and they refused to help me because “they dont understand my VPN client”. Turned out much later that Radius was disabled for my account — but thank God we shared *my* screen (instead of theirs) and blamed it on Linux.
What makes Windows suck is not Windows itself, but the loads of 3rd party snake-oil, MDM and DLP (data loss prevention – yeah…) software layered on top in a modern corporate environment. Also, proprietary VPN clients 😉
Amen! I really wonder how often they can re-invent SSL VPN and wrap a new branding around it. And every time, they boast about a new, more secure VPN solution — while I just configure another OpenSSL variant in my Network manager. Its so sad and hilarious at the same time.
The funniest shit are the “posture checks” — which download 100 mb of binary blobs on windows, but just wave me through when they see my linux client (shout out to the Cisco SSL idiots).
Uh, yes, proprietary VPN clients big annoyance.
But Windows TCP stack implementation is a bit rigid.
AnAmigian,
There are so many VPNs and they all end up doing the exact same thing. Many of forks of openvpn, and if your lucky they’re even compatible with it, but sometimes they’re not. All the duplicate effort for both users and developers is quite something. It’s worth asking why we’ve failed to consolidate them. I understand why people haven’t embraced Microsoft’s VPN solution. It’s IPSEC based, which doesn’t play nicely on all networks and can often be blocked when traveling when people need it the most. Linux also threw their own VPN implementation into the foray, because why not add another. It’s creates more fun for contractors who get to enjoy connecting to client networks with dozens of VPNs – yippee.
(I’m broken)
Respectfully, I feel like this is a bit of an unfair statement because WireGuard is a serious improvement regarding Code size and Audibility, Speed/Performance, Instant connection and handover. Also much easier to configure.
If I had to setup a VPN, I certainly chose WireGuard before anything else.
Andreas Reichel,
Yeah, everyone has their own reason to justify reinventing the wheel, but I find that it’s often a case of “not invented here”. The world really didn’t need another VPN standard. Linux could have implemented an existing standard in the kernel and I honestly think everyone would have been happy with that. We’d get the kernel performance benefits on linux while still being interoperable with our android/ios/BSD/windows/etc friends. Of course I happen to be in the linux camp anyway, but I can still recognize that we have been responsible for fragmentation across many projects stemming from us just not being good at cooperating with others.
@Alfman,
I very much by your argument in general, but not in this particular case.
Its security related an so readable code and understandable configuration actually matters. Have you looked at IPSEC (I know you did), it is hard to defend from a practical point of view. Also, Linux supports IPSEC of course. It just offers a much needed better alternative.
And just to prove my point: I am sure we both agree that IPSEC is real one while OpenVPN/SSL VPN is just a cheap charly. Yet, you want see much IPSEC in the wild anymore. All my clients have switched to SSL based VPN solutions.
Andreas Reichel
I won’t agree that linux needed yet another incompatible VPN standard to achieve their goal of creating their own kernel implementation. VPNs all have to solve the exact same problems, usually with the same crypto….it’s just different containers. Of course linux kernel devs are free to reinvent the wheel like everyone else, but if we are to criticize there being too many VPN, then I think it’s fair to say linux deserves blame like anyone else. A lot of linux devs have a case of “not invented here”. To be fair though, I could be a hypocrite since sometimes I’ve got it too.
I’m not clear about what you mean by this. I looked up cheap charlie, and it didn’t really clear things up. Why isn’t OpenVPN just as real as IPSEC? I actually like openvpn because it’s portable and works on standard ports.
I mostly used IPSEC when I was a windows admin, but it always had a lot of routing problems. While the fault largely lies with modern ISPs and Wifi providers blocking protocols, Regardless of the reason though it makes a lot of sense that it’s been left behind in favor of VPNs that are less likely to be blocked by NAT routers & firewalls.
I said “routing problems”, but I should have said packet forwarding programs on NAT routers that like to pretend there’s nothing beyond TCP/UDP. that needs forwarding.
IPSec operates at L3 while the SSL/TLS VPN clientless portal variant exposes a full web application to the internet. That web front end is a much larger and messier attack surface than an IKE listener. Enterprise SSL VPN appliances (Fortinet, Pulse/Ivanti, Citrix, Cisco) have been among the most aggressively exploited devices on the internet for years, with a steady stream of pre-auth RCE, path traversal, and auth-bypass CVEs.
With a SSL VPN client, it runs as a privileged process and it parses server-supplied configuration. All sources of local privilege escalation and, occasionally, remote issues. AnyConnect and FortiClient both have histories of local-privesc CVEs in their helper services.
Ranking the smallest-attack-surface-first: WireGuard, then IKEv2/IPSec, then client SSL VPN with the portal disabled, then clientless SSL VPN a distance back. (And this still ignores the additional security gain from clean auditable code and simple configuration.)
Andreas Reichel,
I’m curious why you would say this? Honestly I don’t see a reason that opening a TCP/UDP socket exposes your application to any more than opening up a custom L3 protocol. TCP/UDP are very mature now and have robust and secure APIs.
In terms of userspace versus kernel space VPN I’d argue that implementing the VPN as a kernel module opens up a much greater attack surface. I’m ok with implementing a kernel level VPN, however can you agree that it goes against the “microkernels offer improved security and isolation” principal?
Sure VPN appliances can have vulnerabilities, but I’m talking about the VPN tunnels themselves.
This may be the case on windows, I haven’t checked, but it’s not the case on linux. An unprivileged process can access both the tunnel and socket. You may need privilege to create iptables rules and configure routing, however the VPN itself can run as an isolated user without root privileges.
Honestly I think you are wrong, VPNs have a higher attack surface in the kernel than a userspace VPN, potentially making the stakes significantly worse for wireguard if a vulnerability were found.
In any case though, this is all tangential to our original question: whether linux could have used an existing VPN protocol over the wire instead of creating one more. It seems to me that all of wireguard’s users today would have been just as happy using an existing standard over the wire.
@Alfman,
thank you for your time and I don’t want to drag this forever. My view is: Impact on breach vs. sheer possibilities to breach.
Of course we agree, that once you breach a kernel module/instance, its over.
However, that is exactly my point: Wireguard consists of 4k lines well understood code and simple configuration. In my understanding this narrows down the attack vector considerably.
But I am fine when we can not agree here. Best and cheers.
You sound like a sick doctor. It’s said that they are annoying patients. ;-D
And the emoji picker and clipboard history aren’t even new features in Windows 11. They are in Windows 10 too.
Seeing the fastfetch output on Windows ist just cursed…
as is a terminal with a white background.
while windows has plenty of flaws, the sorry state of explorer – as you pointed out – being one of them, i do feel some remarks could use some context.
keyboard layout: i first read your review of the laptop you’re using where you mention you went for an ansi layout instead of the superiour iso option. standard ansi does not have alt gr, so not having a no dead keys alt gr option in the default keymaps is expected. upon closer inspection your keyboard also doesn’t seem to have a menu key? as an “azerty be-nl” layout user i do see how annoying not having alt gr working as expected must be. (side note: openbsd also lacks “us – nodeadkeys on alt gr” on the console)
fans: your review mentions the laptop has fan adjustment options in the firmware. are any windows drivers provided for the fan features? how do you know if the fans are controlled by the firmware or the os at a given time?
touchpad gestures: also likely driver issue. does the pad use a common chipset and/or has it’s pci id been modified?
software package management: stuff from the windows store is pretty well contained. all the rest, well, it’s the wild west. then again, that’s also how i feel about linux software management. why can’t i use an opensuse rpm on redhat? (no need to explain, i know why – this is just an example)
microsoft ecosystem: brave to go all in on edge & outlook 😀 . as a counterpoint, do you also go all in on linux desktop environments? if you use kde, do you also use kmail and konqueror. or if you use gnome, evolution and gnome web?
consistent ui: strangly enough this also happens on linux: my eterm and urxvt seem to have nothing in common, qt & gtk don’t care about each other nor the 2 terminals i mentioned, and what’s up with tk & fltk applications doing their own thing? my pet peeve: gnu getopt allowing long arguments with a single dash instead of only a double dash 😉
none of this takes away of the bigger concerns of privacy, operating system transparency, forced defaults, copilot everything and everywhere, etc….
as said, i felt some remarks lacked a bit of nuance
Pardon? You can say whatever you want about Linux, but package management is certainly one of its superior strengths — for all kind of tastes and flavors. Ebuild, RPMS, Debs, tar.gz, FlatPaks, AUR,
Which other OS/Eco-system provides more or better choices and consistency?
On your question
: because those are dynamically linked binaries with different dependencies? And this problem is solved by RPMFind and friends?
the deluge of options is the problem: the number of formats, package managers, repositories, … the lack of standards where which types of files should go. do config files go in /etc, /usr/etc, some weird place systemd seems to use. do binaries go in /bin, /usr/bin, /usr/local/bin, /opt, ….
if you’re looking for choice i can’t argue, for consistency however: https://man.openbsd.org/hier
dynamic linking isn’t (or should not be) the problem. the most basic failure is simply that a prereq has a slightly different name or versioning scheme. close second is every distro having it’s own idea where which files should go.. a package manager that can’t find the correct shared objects is common – which i don’t get since a dynamic binary just tells you what it needs with `ldd`
Fair and correct assessment, but at least to me, those are all features, not failures. I am a big ARCH/AUR fanboy ( yes, we always need to tell you 🙂 ) and it never let me down especially since I can’t afford easily to re-install all my desktops at various places or the laptops for travel.
On your RPM challenge: I understand your frustration, but please understand that some users prefer very conservative distros (with old and defensive library versions) while some users want bleeding edge.
I believe, you make a mistake when you see Linux as one product — instead each (major) distribution providing its own independent product with distinct features.
That’s why I worked on an alternative way to “launch” applications by creating an “online repository” (I add applications I use from time to time) :
https://github.com/wenuam/wm_app
Andreas Reichel,
I’ll be honest, this is one of my critiques of linux distros. In principal one should be able to install the exact version(s) one wants/needs independently from the installed distro. Yet linux makes this extremely difficult. Short of avoiding the repos and using an alternate package manager that offers more flexibility, the choice between “bleeding edge” versus “conservative” is generally all or nothing. IMHO this shouldn’t be the case and is a notorious shortcoming for many linux distros.
Linux repos were extremely innovative, but they ropped us into a lot of complexity and technical baggage. I know there are reasons behind the complexity (ie dependency hell), but I still wish that linux packages could be significantly simpler and more portable. This will probably sound crazy, but I kind of miss the way DOS software (and early windows software) was packaged. All you needed was to unzip it and it worked without any fuss whatsoever (at least as it relates to packaging).
That’s a fair point, although I think it dismisses (rather than addresses) the frustrations that inphobia is talking about.
Speaking from the perspective of a professional computer user (rather than someone that uses a computer as a professional spreadsheet operator or professional typesetter or web browser user), I just gave up on the idea of running my life from a single OS.
I also gave Windows 11 an honest shot, because it is the only operating system that covers all my use cases, from driving ancient film scanners and sound cards to very specialized photo editing software and my arduino fun. ReFS is fine, finally. GPU support is top notch. It’s been stable. But:
– no matter how many TBs of RAM you have, it is still slow. Everything lags a bit, even if you disable all animations.
– sometimes you are on battery, 10h remaining and, because you are just “taking your time” or doing something light, Windows decides it is a good time to run whatever garbage crap it does and suddenly you are stuck with 3h battery life remaining (battery saver mode helps nothing)
– it slowly drifts into chaos: appdata folder, user profile folder, etc, applications always leave garbage behind.
Then FreeBSD is amazing. I get 6h battery life on an old ThinkPad W530 with a battery that is at 45% capacity while doing real work, but support for newer laptops is still quite deficient. So my ThinkPad P1 runs Linux dual booting Windows, and sometimes I use VMs.
If I were a web terminal user like most people, I’d daily-drive Haiku.
Great writeup.
Interesting how many of the issues you’ve highlighted, I’ve gotten used to as the “norm” as a Windows user.
And thanks for mentioning the clipboard Win+V shortcut – I never knew that existed, and its great!
Uh, I just checked, I double click on zip or tar.gz file and I’m inside of it.
Might it be caused by use of PowerToys?
sobkas,
I thought this was the case as well. But I wasn’t sure because it’s possible that it was an extension rather than vanilla windows. When you stop daily driving windows, these details fade into the background, haha.
current w11 25h2 also handles 7z & several other compression formats. however the review said “Dolphin and Nautilus handle compressed files entirely transparently”. my understanding of transparent would be that an archive behaves exactly like a regular directory, which is not the case on windows in my view, opening != transparent use.
inphobia,
I tested this just now and my copy of windows 11 pro opens zip files inside of windows explorer transparently, just like a directory.
This makes me wonder if Thom might have installed another archive program that replaced windows explorer file associations? When I look at the file associations for ZIP, it’s set to “windows explorer”. AFAIK this isn’t a new windows feature and I’d expect it to work on Thom’s machine as well. I’m only guessing here, but perhaps it got set to something else on Thom’s machine? It seems conceivable that another archive program installed out of necessity or habit could have changed the default file associations.
I imagine we could get to the bottom of it by exchanging screen shots but I kind of doubt that Thom cares to fix it now.
I’m not talking about *opening* a compressed archive; I’m specifically talking about *extracting* it. These are two entirely different things.
Thom Holwerda,
Thank you for the clarification.
I’m looking now at the extraction options available without opening the archive under Dolphin and Windows explorer…
I didn’t bother taking screen shots because I hope this is good enough. KDE’s “Extract Archive To…” is functionally equivalent to window’s “Extract All…”. Both allow the user to select a directory and click extract. The option is one menu deeper in KDE, so selecting the menu item could be a tad quicker under windows, but the tradeoff is that windows does not have the other two menu options. Although they merely perform the same operation as “Extract Archive To…” with a default path.
(I’m not in front of a gnome desktop right now, are they much different?)
Are these the extraction options you are referring to or has something once again gone over my head? Haha.
Edit: I just learned that UniGetUI is now also available on Linux and Mac.
Frankly, I would like to see a UniGetUI equivalent on Linux an I know of none.
PackageKit thankfully handles all of the distro specific package managers: Zypper, DNF, Yum, Apt, pacman, etc. .
And KDE Discover then handles PackageKit, Flatpak, FwUpd, and Plasma Addons.
But UniGetUI handles much more.
I thought it was weird “US (int’l with AltGr dead keys)” didn’t work the way I expected. I’m used to the Windows US Intl behavior, and even made my own expanded US Intl layout in MSKLC.
Or try perhaps Cfilorux (also done with MSKLC) : https://github.com/wenuam/wm_key_map_cfilorux
Alphabetical in columns, with many many dead keys for most of EU languages.
here’s mine, if anyone wants to try it.
https://github.com/SarreqTeryx/usiplus
Since I’m a former Windows user who now tries to avoid using it, I kind of anticipated what would be in this review. But I still find it cathartic. These days, I split my desktop time between, Fedora Silverblue, Fedora COSMIC Atomic, and a Chromebook, all of which I enjoy, but for slightly different reasons and use cases.
Good writeup. I’m in the camp that actually likes the Windows 11 UI (it looks much more polished than 10 to me), but they really have regressed the basic behaviour of Windows in the last few years. As you say, the File Explorer is really slow. Shoving Copilot into basic utilities like Notepad is crazy. And the refusal to work with users that want a local account is a dealbreaker for me. The workarounds for that are getting a bit painful.
Paradroid,
It did get slow.. Not too long ago I was tasked with installing software on user machines and the bad performance stuck with me. What in the world is windows doing? It can be slow even when I don’t see a reason for it to be slow. As always the answer is to ignore the overhead and buy better hardware to make it faster haha.
Executives decided that windows users are to be forcefully tethered to microsoft accounts & servers whether they want it or not 🙁
The older I get, the more I absolutely despise this “throw better hardware at it” solution to software bloat (and to be clear, I’m not directing this at you, I’m sure you share my frustration). We live in a time when a mini PC the size of a deck of cards with a six watt TDP CPU can outperform a 125W TDP monster CPU based build from 2010, yet Windows file manager on a maxed out 2026 Threadripper Pro workstation still takes several seconds to open, as if we were still on that 2010 beast.
A couple of the lightweight file explorers I like to use on Windows are Explorer++ (when run as admin, very handy for copying files without taking ownership, portability a plus) and Q-Dir (great UI for reorganising files between multiple locations but has plenty of other useful little features), both are quite snappy.
rhy7s,
Thanks for the suggestions. I expect most of us are actually running linux/bsd ourselves, so it’s more of a curiosity than anything. Given that modern computers are so blazing fast, it’s peculiar that windows explorer can be laggy with simple functions. I’m just guessing here, but it’s got to be doing something right? MS security essentials? MS spyware/AI scanning? One drive?
I’m sure somebody can profile it and figure out the culprit, but this time it won’t be me because I can’t be bothered 🙂
I use Explorer++ as my runas access. Nice. Ah, I remember Q-Dir. Many views.
But my main file browser is the heavy gun Directory Opus for Windows. ☺️