The Windows Subsystem for Linux is now open source

I can tell that if POSIX subsystem in Windows NT was actually a good enough UNIX experience, I would never bothered with those Slackware 2.0 install disks.

And the subsystems concept was quite common in micro-computers and mainframes space, Microsoft did not come up with the idea for Windows.

WSL 1 works fine. I much prefer it over 2 because I only run windows in a VM and nested virtualization support isn't all there.

Also feels a lot less intrusive for light terminal work.

That would not be unique, as is what BSD has done for Linux compatibility basically forever.

Performance was one problem, but imho the biggest was that MMAP semantics were inherited from the NT side and made a lot of applications crash (mmap's created could only be as large as the file's current size as in Windows, while Linux/BSD semantics allows for a mmap larger than the file that's usable without constant remapping as the file grows).

They didn't prioritize it until fixing at a late stage, barely before WSL 2 came out. Sometimes i do wonder if they made a premature decision to move to WSL2 since there was quite a lot of basic applications/runtimes that were crashing due to this fix lacking (Naturally a lot of other new Linux API's like io_uring probably would have made it an api chasing treadmill that they just wanted to circumvent).

Win32 APIs like CreateProcess suck because they have to spend so much time setting up the stuff that allows Win32's application model to mimic that of 16-bit Windows, which was coopreratively multitasked. The NT kernel is much faster at creating processes when it doesn't need to worry about that stuff.

As for NTFS: it's not NTFS specifically, it's the way the I/O system is designed in the NT kernel. Imagine any call from outside that layer transitioning through a stack of filter drivers before actually reaching the implementation. Very powerful stuff, but also very bad for performance.

Hm. I used Git on WSL1 for many years, with medium sized repos hosted on a Windows drive, and it worked great. When I moved to WSL2 Git became a whole lot slower - it now takes about 5-8 secs to execute 'git status' where before it was instant.

Windows actually created a new process type for this: Pico processes[1]. This allows WSL1 to perform quite a bit better than Cygwin on something like Windows XP.

1. https://learn.microsoft.com/en-us/archive/blogs/wsl/pico-pro...

>But WSL2 is freaking incredible

It's good. But if/when you start using it as your main work platform nagging issues start cropping up. The native linux filesystem inside it cannot actually reclaim space. This isn't very noticeable if you aren't doing intensive things in it, or if you are using it as a throwaway test bed. But if you are really using it, you have to do things like zero out a bunch of space on the WSL disk and then compact it from outside in the Windows OS. Using space from your NTFS partition / drive isn't very usable, the performance is horrible and you can't do things like put your docker graph root in there as it is incompatible. It also doesn't respect capitalization or permissions and I've had to troubleshoot very subtle bugs because of that. Another issue is raw network and device access, it basically isn't possible. Some of these things are likely beyond the intended use of WSL2, in its defense. Just be aware before you start heavily investing your workflow in it. For these use cases a traditional dual boot will work far better and save you much frustration.

Why not just use Linux then?

The whole point of Windows right now is having a kernel that a) does not shove GPL down the device manufacturer's throat and b) care about driver API stability so that drivers actually work without manufacturer or maintaner intervention every kernel upgrade.

> I was never a fan of powershell, sh/dash/ash/bash seem fine

It depends on what you're doing. PowerShell is incredible for Windows sysadmin, and the way it pipes objects between commands rather than text makes it really easy to compose pretty advanced operations.

However, if I'm doing text manipulation, wrangling logs, etc, then yes, absolutely I'm going to use a *nix scripting language.

I sometimes say, tongue in cheek slightly, that the best Linux desktop is Windows.

WSL 2 runs a full Linux kernel under Hyper-V. There are some out-of-tree or staging drivers included in Microsoft's Linux kernel derivative and they publish their kernel sources at https://github.com/microsoft/WSL2-Linux-Kernel.

Note that in recent versions of Windows, typically the bulk of Windows now runs under a hypervisor (i.e., "in a VM") as well: https://learn.microsoft.com/en-us/windows-hardware/design/de...

It’s finally enough for me at least. I’m skipping windows 11 and going to Linux instead.

When we die off I guess lol.

I've been using windows since I was 6 or 7. I currently work in a Mac environment and hate it. I worked in a linux one for 5 years. Nothing feels like the first language you learned I guess?

My home computer is windows and it'll be that way until windows stops existing.

Edit: when I say we I mean the people still on windows.

I would be very happy with Windows 7 on manufacturing side - lots of CNCs that are still in use and supported by manufacturers are still on Windows 98.

The higher up have such a hardon for Microsoft, I think it could actually be used as a bridge across the Atlantic ocean. We've already spent years migrating shit off of microsoft platforms onto the newest and latest microsoft platforms.

I left some room for myself with "a good reason" :)

When company is forcing you to use something out of inertia, then it's probably not for a good reason.

Actually regarding the "global scale" – I'm not really sure it's true, I think MS has influence mostly in US. Many EU and Asian companies I worked with were using OSX/Linux.

Yeah, I totally agree with what's being said here. It's a tough pill to swallow when you realize just how entrenched Microsoft is in the business world, and how difficult it would be to get everyone to make the switch to Linux.

I mean, think about it - most companies are still stuck on Windows 10 or 11, and they're using all those Microsoft services like Teams, Outlook, and Exchange. It's like they're trapped in this Microsoft ecosystem, and it's gonna take a lot more than just a few people saying "hey, let's switch to Linux" to get them out of it.

And don't even get me started on the IT departments in these places. A lot of them are just kids who memorized some CompTIA exams and don't really know what they're doing. They're using Windows 11 at home, but they have no idea how to deal with all the outdated Windows 7 machines that are still being used in manufacturing.

Microsoft, on the other hand, has been really smart about this. They've managed to get their products and services woven into the fabric of how we do business on a global scale. It's gonna take a lot more than just a few open-source projects to change that.

You're saying it like there is no alternative and you can't just open and edit same excel files in Libre Office Calc, Google Sheets or Numbers without any problem whatsoever.

AAA and AAAA games became (expensive) gateways to microtransaction based money extraction application, in my opinion.

I enjoy older, smaller games nonproportionately more when compared to big titles which require much more resources and time. Yes they look nice, yes they use every documented and undocumented feature of my GPU, yes "it's so fluffy", but it is not enjoyable, esp. with shoved down microtransactions.

If we're talking FPS, give me any Half-Life (and Portal) title and I'm good. Gameplay first, unique art direction, good story, and a well built universe which is almost palpable with lore.

If we're talking RTS, C&C series, Dune Emperor, Supreme Commander and StarCraft is enough.

I've been using Arch for 1-3 years now, as far as I can remember the only time that my system "break" was caused by pacman lock got stuck somehow. Aside of that it's pretty stable in general.

If you don't have an adaptive sync +variable refresh rate) monitor and everything set up to use it, and don't like screen tearing (you enable vsync wait), overrunning the frame budget (e.g 16ms for 60hz) can mean dropping down to half the frame rate.

But I'm hunting for reasons here. A gaming setup should be using adaptive sync so those concerns mostly go away. But there may be problems with Linux support.

Don't get me wrong, what I meant is that I only uses windows on games that runs poorly for me, I use Linux as my daily driver.

Regarding fps, it's around 15fps diff, and it's bad in my case because I had a potato machine.

There are even games like Infinity Nikki with anti-cheat that allows the Steam Deck but specifically detects and blocks desktop Linux. You have to wonder if that gets them any real security since the method they use to detect the Deck is probably spoofable.

There is more nuance to the anti-cheat systems supporting Linux argument than "it supports it but they won't use it". Turning on Linux support does weaken the security posture of the anti-cheat system, so it's not simply a decision of "it works with Linux, but they won't do it". It is moreso a question of whether the security posture changes for the game with this platform support enabled meet the business requirements. It's not a surprise that games with high MTX revenue do not turn this on, as I imagine this would be the biggest concern with this weaker security posture.

One of the boons of console hardware is also the strict execution environment that is presented on the system. While this of course doesn't prevent all cheating behavior in online games, a large selling point of it as a platform to publishers is not only the market segment available, but the security aspects of the runtime environment.

Really hope valve’s server side anti-cheat will be a success and more competitive games will move over to that.

Nowhere it says they don't want to.

AoE2:DE has a gold rating, but multiplayer doesn't work at all, and it's not even due to anticheat.

Denuvo counts each proton version as a unique activation, might help you avoid this issue going forward

Can you remember any particular problems in Overwatch? I've been down that road, so there's a chance I might have some info that you would find useful.

One problem that was unsolved last time I checked: Saving highlight videos. It used to work if you told Overwatch to use webm format instead of mp4, but Blizzard broke that somewhere along the line, possibly in the transition to Overwatch 2. (I worked around this with OBS Studio and its replay buffer feature.)

In general you want to avoid Nvidia if you want to play games on Linux, but maybe things will get better.

i play rdr1 via proton on opensuse + amd and i get better frames than windows

> You can do very few productive things with Windows other than software development.

I guess you meant Linux here

A fair comment, but the argument I'd make against that is a lot of those creative tools are moving to the web. I personally work for Figma, and have seen that first hand. UI/UX design was entirely OSX/Windows centric for the last 40 years, and now it's platform agnostic. Even video editors are just at the nacent stage of looking at the web as an editor surface.

Totally hear you though for things like CNC milling software that's meant to stay static for the lifetime of the mill - that's not going anywhere.

And this is why wine/proton are so good: they’re implementing the only defacto stable API that exists.

Yeah, I really like my Mac, but third-party software isn't its strong suit. It's hilarious how often Apple will wholesale break like half the software in existence.

>And on top of that, Linux is completely unusable to Japanese/Chinese speakers due to how hard it is to input the moon runes

How do Deepin and such solve this?

Linux on HN is always an example of https://xkcd.com/2501/

How many months can you use a Linux desktop to do daily externally mandated processes and not drop down to a bash shell at some point?

Average consumers and users do not want to use the unix utilities that Linux people love so much. Hell, developers barely want to use classic unix utilities to solve problems.

Users do not know what a "mount point" is. Users do not want a case sensitive file system. Users do not want an OOM killer that solves a poor design choice by randomly culling important applications at high utilization.

Users do not care for something that was designed in the 60s before we understood things like interface design and refuses to update or improve due to some weird insistence on unix purity.

Users do not care about ABI stability. They care about using the apps they need to use. That means your platform has to be very easy to support, Linux is not at all easy to support, and at least part of that is a weird entitlement Linux users feel and demonstrate in your support queue.

Hilariously, users DO WANT a centralized app repository for most day to day apps! Linux had this forever, though it had mediocre ergonomics and it was way too easy for an average computer user to manage to nuke their system as Linus Sebastian found out in a very unfortunate timing situation. Linux never managed to turn this potential victory into anything meaningful, because you often had to drop into a bash shell to fix, undo, modify, or whatever an install!

On linux it's quite trivial. KVM is part of the kernel. Installing libvirt and virt-manager makes it really easy to create vms.

I'd say even passing through a GPU is not that hard these days though maybe that depends on hardware configuration more.

What 3D modelling were you doing that couldn't be done on linux?

Sadly I'm not one of those people because I have a desktop with an AMD Ryzen 7 5800X3D, which does not have an integrated graphics card.

However now that AMD is including integrated GPUs on every AM5 consumer CPU (if I'm not mistaken?), maybe VMs with passthrough will be more common, without requiring people to spend a lot of money buying a secondary GPU.

There is also native context for VirtIO, but for now Windows support is still not planned.

Also note some brave soul implemented 3D support on KVM for Windows. Still in the works and WinUI apps crash for some reason.

True, but I don't have the need to run applications that require GPU under WSL, while I do need to run applications that require the GPU under my current host OS. (and those applications do not run under Linux)

There are computers that size, but I guess you mean with a male PCIe plug on them?

If the card is running its own OS, what's the benefit of combining them that way? A high speed networking link will get you similar results and is flexible and cheap.

If the card isn't running its own OS, it's much easier to put all the CPU cores in the same socket. And the demand for both x86 and Arm cores at the same time is not very high.

You may be interested in SmartNICs/DPUs. They're essentially NICs with an on-board full computer. NVIDIA makes an ARM DPU line, and you can pick up the older gen BlueField 2's on eBay for about $400.

> full fledged computers in a GPU sized package

.. isn't this just a laptop or a NUC? Isn't there a massive disadvantage in having to share a case or god forbid a PCIe bus with another computer?

It's really nice if you have two seperate GPUs in your computer?

Last I looked, Thunderbird used Exchange Web Services to connect to Office365 which Microsoft is getting rid of: https://techcommunity.microsoft.com/blog/exchange/retirement... (I point out Office365 since vast majority of "Exchange" users are on 365)

It also only support email and not calendaring/contacts.

That being said, Office365 Web Client is pretty good at this point and someone who doesn't live in Office all day can probably get along fine with it.

Yes but the point is to not do dev on windows, because window "terminal" sucks

I don't do anything like USB pass through. I use it purely for the linux tooling to support software development. I find it a perfectly suitable system for that purpose and would recommend it.

there is GUI program to automatic this progress[1]

I have been using WSL to develop Firmware in Zephyr, no problem so far.

[1]: https://blog.golioth.io/usb-support-in-wsl2-now-with-a-gui/

You can make your own clean version, legally, with this file. https://schneegans.de/windows/unattend-generator.

I get customers and most people don't know about it but it's kind of ridiculous that techy people in a tech forum don't know how to do it.

There is no flavor of Windows 11 that is acceptable. Even the UI itself is a disaster. A cornucopia of libraries and paradigms from React Native to legacy APIs as if an interdimensional wave function of bad ideas had collapsed into an OS, but with ads.

I use Windows 11 IoT Enterprise LTSC, which as far as I'm aware has all the features that Pro has (plus the IoT Enterprise stuff) and zero bloat. I switched to it from my already de-bloated 11 Pro installation (because it removes some telemetry you're normally unable to disable) and have had 0 issues with it. I can't say I activated it using a normal retail product key, however, there are easy solutions to that.

Ya I totally get that. The way I view it is that windows is a glorified driver support layer and any actual work i do is almost exclusively in the Linux container.

When I used to have free time it was great for games too

I replaced my m1 with a snapdragon laptop running Win11 and upgraded that to pro. For what I do with it, it runs great with very long battery times, for less than Apple quoted to repair the m1. I don't use the copilot features and haven't seen any ads so far, except maybe for office during setup.

Maybe just for your specific preferences. Terminal is plenty fine. Vertical taskbar on the right is straight up user preference. PowerToys for an app starter? Like Alfred? The start search does a decent enough job of that. Folder Size is nice, but enumerating all files is very taxing.

Oh running Ice to wrangle the menu bar app icons or Rectangle to properly manage windows ('cause Apple screwed that one up) must be unnecessary.

Each OS is going to have extension applications to improve on the OOTB experience. This is an invalid argument to choosing one over the other.

>Windhawk for a vertical taskbar on the right

Huh? Windows supports vertical taskbar.

"I find it dismaying that people on Hacker News willingly submit to incredibly user-hostile behavior from Microsoft"

And I find it funny that the crowd that spends whole days implementing user-hostile features in yet another SaaS crapware has so much to say about Microsoft's bad behavior.

There is an additional reason: Some (many?) people simply prefer the Windows UI conventions (once you remove all the enshittifications post Windows 7).

Only on countries where people earn salaries big enough to pay for the Apple hardware tax.

Superior hardware with terrible software. Also they straight up artificially limit their hardware so they don't cannibalize their sales, which is slightly understandable, but they do it in the dumbest ways. My SOs MacBook Air can only do one external monitor, even though it has the same specs as her work Pro. Oh and good luck actually getting that external display to work, I swear only like 50% of USB-C docks work on the platform.

>Macbooks are winning the laptop war because of superior hardware.

No. This is just you repeating marketing.

No Nvidia chip = B tier at best.

I have a $700 Asus with a 3060 that is better. Go ahead and scale up to a $2000 computer with an Nvidia chip and its so obviously better, there is nothing to debate.

No one cares about performance per watt, its like someone ran a 5k race, came in 3rd and said "Well at least I burned fewer calories than the winner!"

Some of them are.

But the increasing market share of Macs and even Linux these days plus the ever increasing of OSS initiatives from Microsoft points out that Microsoft knows a lot fewer of their users are as captive as they were in the 90's, for example.

My development is mainly Windows and I prefer Linux host with Windows VM guests. The experience is more stable and I can revert to a snapshot when Windows or Microsoft product update brakes something or new test configuration does. It also allows to backup and retain multiple QA environments that are rarely used, like a client's Oracle DB. It is nice being able to save the VM state at the end of the week and shut it all down so you can start the next right where you left off. Cannot do that when your development environment is the bare metal OS. Windows has known issues of waking a sleeping laptop.

In the same spirit if "it depends", there are other options that may work for people with different Linux/Windows balance points:

* Wine is surprisingly good these days for a lot of software. If you only have an app or two that need Windows it is probably worth trying Wine to see if it meets your needs.

* Similarly, if gaming is your thing Valve has made enormous strides in getting the majority of games to work flawlessly on Linux.

* If neither of the above are good enough, dual booting is nearly painless these days, with easy setup and fast boot times across both OSes. I have grub set to boot Linux by default but give me a few seconds to pick Windows instead if I need to do one of the few things that I actually use Windows for.

Which you go for really depends on your ratio of Linux to Windows usage and whether you regularly need to mix the two.

> Forscan for Ford vehicles also has to run under Windows.

I've successfully run it with WINE. Thought, my Forscan executable was 3 years old or so and that may have changed, but I doubt it.

> For example, I use Solidworks, so I need to run windows.

Right. One of the things a lot of people don't get is the extent to which multidisciplinary workflows require Windows. This is particularly true of web-centric software engineers who simply do not have any exposure to the rest of the engineering universe.

Years ago this was the reason we had to drop using Raspberry Pi's little embedded microcontroller. The company is Linux-centric to such an extent that they simply could not comprehend how telling someone "Just switch to Linux" is in a range between impossible and nonsensical. They were, effectively, asking people to upend their PLM process just for the sake of using a little $0.50 part. You would have to do things like store entire OS images and configurations just to be able to reconstruct and maintain a design iteration from a few years ago.

WSL2 is pretty good. We still haven't fully integrated this into PLM workflows though. That said, what we've done on our machines was to install a separate SSD for WSL2. With that in place, backing-up and maintaining Linux distributions or distributions created in support of a project is much, much easier. This, effectively, in some ways, isolates WSL2 distributions from Windows. I can clone that drive and move it from a Windows 10 machine to a Windows 11 machine and life is good.

For AI workflows with NVIDIA GPU's WSL2 is less than ideal. I don't know if things have changed in this domain since I last looked. Our conclusion from a while back was that, if you have to do AI with the usual toolchains, you need to be on a machine running Linux natively rather than a VM running under Windows. It would be fantastic if this changed and one could run AI workflows on WSL2 without CUDA and other issues. Like I said, I have not checked in probably a year, maybe things are better now?

EDIT: The other reality is that one can have a nice powerful Linux machine next to the Windows box and simply SSH into it to work. Most good IDE's these days support remote development as well. If you are doing something serious, this is probably the best setup. This is what we do.

Did you know that Forscan works flawlessly under Wine if you're not using Bluetooth?

I really want to like Windows 11, and I enjoy using WSL, but Microsoft treats me too much like an adversary for me to tolerate it as a daily driver. Only a complete scumbag of a product manager would think pushing Candy Crush ads is a good idea.

I’ve got an airgapped Toughbook that I use for the few Windows apps I really need to talk to strange hardware.

I run Windows in a VM where I need windows. It’s so much easier to fix a broken Linux installation than a broken Windows installation.

Saying running full Linux avoids wasting time on fiddly workarounds kinda blows my mind.

Full hardware support is still not a given, and Windows emulation is still need for so many cases (e.g. games, specialized software etc).

Until I can choose any machine based on form factor and specs alone and just run Linux on it, WSL will the best version of Linux it can run.

What detects it is running on windows out of interest?

I use WSL extensively, with lots of languages, and I’ve never had anything do that.

It’s running in a VM, so that would be some kind of weird VM escape?

I would love to hear about these edge cases and which tooling fails to detect they they were launched form linux.

Sounds a lot like a picnic problem but you didn’t give nearly enough details.

If on WSL2, they need

[interop]

appendWindowsPath=false

section in /etc/wsl.conf.

Then everything will go flawlessly.

For me, the best part of running Linux as the base OS is not having to deal with Windows.

No ridiculous start menu spam; a sane, non-bloated operating system (imagine being able to update user space libraries without a reboot, due to being able to delete files that other processes still have opened!); being able to back up my data at the file level without relying on weird block-level imaging shenanigans and so much more.

How is inverting the host/guest relationship an improvement on that?

But you still get the worst of the Windows world, which is more than many are willing to deal with. I was using windows for years as my main gaming OS, but after they announced W11 being the only way forward. Switching to Linux on the desktop was like a breath of fresh air. I'll leave it at that.

If I were to run an OS on a VM it's gonna be windows, not Linux

> You get the best of both worlds with WSL.

You obviously don't. Maybe WSL is the best compromise for people who need both Windows and Linux.

But it's ridiculous to think that WSL is better than just Linux for people who don't need Windows at all. And that's kind of what the author of this thread seems to imply.

Similarly powerful would be totally fine. More powerful really is silly. Personally I couldn't make a lot of my workflows work very well with WSL2. Some of the stuff I run is very memory intensive and the behavior is pretty bad for this in WSL2. Their Wayland compositor is also pretty buggy and unpolished last I used it, and I was never able to get hardware acceleration working right even with the special drivers installed, but hopefully they've made some progress on that front.

Having Windows and Linux in the same desktop the way that WSL2 does obviously means that it does add a lot of value, but what you get in the box isn't exactly the same as the thing running natively. Rather than a strict superset or strict subset, it's a bit more like a Venn diagram of strengths.

As everyone said, WSL2 is actually virtual machines and it is what most people are actually using now. That said, I feel the need to chime in and say I actually love WSL1 and I love Windows NT the kernel. It bums me out all the time that we probably won't get major portions of the NT kernel, even an out-of-date version, in some open source form.

I like Linux, and I use Linux as my daily desktop, but it's not because I think Linux or even UNIX is really that elegant. If I had to pick a favorite design it would be Windows NT for sure, even with all its warts. That said, the company behind Windows NT really likes to pile a lot of shit I hate on top of that pretty neat OS design, and now it's full of dubious practices. Automatic "malware submission" on by default, sending apps you download and compile yourself to Microsoft and even executing them in a VM. Forced updates with versions that expire. Unbelievable volumes of network traffic, exfiltrating untold amounts of data from your local machine to Microsoft. Ads and unwanted news all over the UI. Increasing insistence in using a Microsoft account. I could go on and on.

From a technical standpoint I do not think the Linux OS design is superior. I think Linux has some amazing tools and APIs. dmabufs are sweet. Namespaces and cgroups are cool. BPF and it's various integrations are borderline insane. But at its core, ... It's kinda ugly. These things don't all compose nicely and the kernel is an enormous hard-to-tame beast. Windows NT has its design warts too, all over, like the amount of involvement the kernel has in the GUI for historical reasons, and the enormous syscall surface area, and untold amounts of legacy cruft. But all in all, I think the core of what they made is really cool, the subsystems concept is super cool, and it is an OS design that has stood up well to time. I also think the PE format is better than ELF and that it is literally better for the capabilities it doesn't have w.r.t. symbols. Sure it's ugly, in part due to the COFF lineage, but it's functionally very well done IMO.

I feel the need to say this because I think I probably came off as a hater, and tbh I'm not even a hater of WSL2. It's not as cool as WSL1 and subsystems and pico processes, but it's very practical and the 9p bridge works way better than it has any right to.

Thanks for pointing this out.

It used to be. They moved to a VM.

Turns out that it's easier to emulate a CPU than syscalls. The CPU churns a lot less, too, which means that once things start working things tend to keep working.

You're thinking of the POSIX personality of Windows NT of old. This was based on Interix and has been deprecated about two decades ago and is now buried so deep that it couldn't be revived.

The new WSL1 uses kernel call translation, like Wine in reverse and WSL2 runs a full blown Linux kernel in a Hyper-V VM. To my knowledge neither of these share anything with the aforementioned POSIX subsystem.

WSL 2 is actually virtualized despite the name

WSL1 was a subsystem. WSL2 is mostly a VM.

They had to give that up because it was too slow, I think for IO. Unfortunate.

I mean... WINE does the same on windows, but microsoft refuses to release their API docs for all internal APIs. They release WSL by relying on Linux's open-ness, while refusing the same for themselves.

It's deeper but let's not overblow it.

I think the two fairly deep integrations are window's ability to navigate WSL's filesystem and wslg's fairly good ability to serve up guis.

The filesystem navigation is something that AFAIK can't easily be replicated. wslg, however, is something that other VMs have and can do. It's a bit of a pain, but doable.

What makes WSL nice is the fact that it feels pretty close to being a native terminal that can launch native application.

I do wish that WSL1 was taken further. My biggest grip with WSL is the fact that it is a VM and thus takes a large memory footprint. It'd be nice if the WSL1 approach panned out and we instead had a nice clean compatibility wrapper over winapi for linux applications.

> The integration between Windows and the WSL VM is far deeper than a typical VM hypervisor.

Sure, but I never claimed otherwise.

> You cannot claim with a straight face that Virtualbox is easier to use.

I also didn't claim that. I wasn't comparing WSL to other virtualization solutions.

WSL2 is cool. Linux doesn't have a tool like WSL2 that manages Linux virtual machines.

The catch 22 is that it doesn't need one. If you want to drop a shell in a virtual environment Linux can do that six ways through Sunday with no hardware VM in sight using the myriad of namespacing technologies available.

So while you don't have WSL2 on Linux, you don't need it. If you just want a ubuntu2204 shell or something, and you want it to magically work, you don't need a huge thing with tons of integration like WSL2. A standalone program can provide all of the functionality.

I have a feeling people might actually be legitimately skeptical. Let me prove this out. I am on NixOS, on a machine that does not have distrobox. It's not even installed, and I don't really have to install it since it's just a simple standalone program. I will do:

    $ nix run nixpkgs#distrobox enter

    $ nix run nixpkgs#distrobox enter
    Error: no such container my-distrobox
    Create it now, out of image registry.fedoraproject.org/fedora-toolbox:latest? [Y/n]: Y
    Creating the container my-distrobox
    Trying to pull registry.fedoraproject.org/fedora-toolbox:latest...
    ...
    0f3de909e96d48bd294d138b1a525a6a22621f38cb775a991974313eda1a4119
    Creating 'my-distrobox' using image registry.fedoraproject.org/fedora-toolbox:latest [ OK ]
    Distrobox 'my-distrobox' successfully created.
    To enter, run:

    distrobox enter my-distrobox

    Starting container...                    [ OK ]
    Installing basic packages...             [ OK ]
    Setting up devpts mounts...              [ OK ]
    Setting up read-only mounts...           [ OK ]
    Setting up read-write mounts...          [ OK ]
    Setting up host's sockets integration... [ OK ]
    Integrating host's themes, icons, fonts... [ OK ]
    Setting up distrobox profile...          [ OK ]
    Setting up sudo...                       [ OK ]
    Setting up user groups...                [ OK ]
    Setting up user's group list...          [ OK ]
    Setting up existing user...              [ OK ]
    Ensuring user's access...                [ OK ]

    Container Setup Complete!
    [john@my-distrobox]~% sudo yum install glxgears
    ...
    Complete!
    [john@my-distrobox]~% glxgears
    Running synchronized to the vertical refresh.  The framerate should be
    approximately the same as the monitor refresh rate.
    302 frames in 5.0 seconds = 60.261 FPS
    ^C

That's just Fedora. WSL can run a lot of distros, including Ubuntu. Of course, you can do the same thing with Distrobox. Is it hard? Let's find out by using Ubuntu 22.04 instead, with console output omitted:

   $ distrobox create --image ubuntu:22.04
   ...
   $ distrobox enter ubuntu-22-04
   ...
   $ sudo apt install openarena
   ...
   $ /usr/games/openarena

On the other hand, even trying pretty hard (and using special drivers) I could never get hardware acceleration for OpenGL working inside of WSL2, so it could be worse.

That aside, everything works. More complex applications (e.g. file browsers, Krita, Blender) work just fine and you get your normal home folder mapped in just like you'd expect.

Yes, yes I can. Also does most of everything. WSL has severe issues with hardware translation.

With WSL you can use “Linux the good parts” (command line tools, efficient-enough paradigms for fork() servers) and completely avoid X Windows, the Wayland death spiral, 100 revisions of Gnome and KDE that not so much reinvent the wheel but instead show us why the wheel is not square or triangular…

I fully agree with you - "YMMV" is the one true take. Visual Studio has never been particularly attractive to me, my whole workflow is filled with POSIX tools, and my code mostly runs on Docker and Linux servers. Windows is just another thing to worry about for me, be it having to deal with the subtle quirks of WSL not running on raw metal or having to deal with running UNIX-first tooling (or finding alternatives) on Windows. If it wasn't for our work provided machines being Windows by default, and at home, being into VR gaming and audio production (mostly commercial plugins), I'd completely ditch Windows in a heartbeat.

Do you believe the 600+ people with the same problem here: https://github.com/microsoft/WSL/issues/4197

WSL 1 had fast IO but couldn't support all features.

WSL 2 supports all features but has famously slow IO.

Example:

1. Shell into WSL

2. Clone a repo

3. Make a bunch of changes to the repo with a program within WSL

4. Run git status (should finish in less than a second)

5. Open repo from a Windows IDE

6. Run git status. This makes windows change each file's permissions, ownership, etc... so it can access the files as git status recursively travels through every file and folder

7. Go for coffee

8. Go for lunch

9. Git status finished after 35 minutes.

10. Close IDE

11. Shell back into WSL

12. Make a change in WSL

13. Run git status from within WSL

14. Wait another 35 minutes as Windows restores each file's ownership and permissions one by one

------------------------------------

The IO overhead is so bad that Microsoft built two new products just to get around it:

1. VSCode WSL remote-client architecture.

VSCode acts as a server within WSL and a client within Windows. Connect both VSCode instances (through proxy/tunnel if needed) and the server can perform the client's File IO ops on behalf of the client rather than letting an Application on Windows try to interact with any of WSL's file systems.

2. Windows DevDrive

Basically set aside a virtual-disk/partition and set it up as a different file system (ReFS) that doesn't use Window's file permissions, ownership and doesn't decrypt then decompress on each file input, doesn't compress then encrypt each file output, and doesn't virus scan the files on usage.

TL;DR Store the files on a network drive and hope race-condition ops from both WSL and Windows don't corrupt any files.

>The problem is Windows IO filters

Not the biggest issue of them, 'find' and 'git status' on WSL2 in a big project is still >100 times slower on windows dev drive which avoids those filters than it is with WSL 1 on dev drive.

WSL 1 on regular ntfs with defender disabled is about 4x slower than WSL1 on dev drive, so that stuff does cause some of it, but WSL2 feels hopelessly slow. And wsl 2 can't share memory as well or take as much advantage of the filesystem cache (doubling it if you use the windows drive in both places I think, unless the network drive representation of it doesn't get cached on the WSL2 drive.

This. WSL was SO much more interesting in v1 times.

I liked the networking in WSL1 more too

I'm guessing they use plan9 because distros already ship support for it, and it's super simple compared to NFS? It doesn't seem like CIFS/NFS would be any faster, and they introduce a lot more complexity.

While I can see the subtle distinction you're trying to draw people's attention to (NTFS is not the problem, filesystem operations generally on Windows are the problem) I have to say it seems like a distinction without a difference in real terms. They made a range of changes that seem to produce more complicated code everywhere because the overhead of various filesystem tasks are substantially higher on this OS vs every other OS.

But in the end they had to get the OS vendor to bless their process name anyway, just so the OS would stop doing things that tank the performance for everybody else doing something similar but who haven't opened a direct line up with the OS vendor and got their process name on a list.

This seems like a pain point for the vendor to fix, rather than everybody shipping software to their OS

I don't know how it's set up. That's kind of my point though. I have to now be an expert in Linux and Windows to debug this stuff, which is a waste of my time as someone who's job it is to develop (server, i.e. Linux) software. I had exactly zero issues when I was using Fedora. At one point my company made all of the Linux users move off (we do now have an IT-supported Linux image, but I haven't found the time to re-set up my laptop and don't fully trust that it will work without a bunch of trouble/IT back-and-forth because they also made Windows users start using passkeys), and since then I've seen way more issues with Windows than Linux (e.g. one day my start menu just stopped reacting to me clicking on programs), in addition to things like ads in the lock screen and popups for some XBox pass thing that I had to turn off, which is just insane in a "professional" OS. A lot of days I end up having to hold down the power button to reboot because it just locks up entirely.

OSX was a bit janky with docker filesystem slowness, homebrew being the generally recommended package manager despite being awful (why do I sometimes tap a cask and sometimes pour a bottle? Don't tell me; I don't care. Just make it be "install". Also, don't take "install" as a cue to go update all of my other programs with incompatible versions without asking), annoying 1+ second animations that you can't turn off that make it so the only reasonable way to use your computer is to never maximize a window (with no tiling support of course), and completely broken external monitor support (text is completely illegible IIRC), but Windows takes jank to another level.

By contrast, I never encounter the issues people complain about on Linux. Bluetooth works fine. Wifi works fine. nVidia GPUs and games work fine. Containers are easy to use because they're natively part of the OS. I prefer Linux exactly because I stopped enjoying "tinkering" with my computer like 10 years ago, and I want it to just quietly work without drawing attention to itself (and because Windows 8 and the flat themes that followed were hideous and I was never going to downgrade to that from Windows 7).

openssh has been an optional windows component for... almost a decade now? including the server, so you can ssh into powershell as easily as into any unix-like. (last time I set it up there was some fiddling with file permissions required for key auth to work, but it does work.)

I dislike using putty, I use the ssh client from WSL. Just feels .. better. And bash/fish history helps.

On the other hand sometimes the GUI on WSL decides to break and you have to restart the whole thing.

Aged like fine milk

Who deals with this? All this is fine out of the box on a modern Linux distro.

You need new reasons to hate Linux, because all those issues were solved a while ago.

I'm confused, in what world does running Linux require more RAM than Windows?

The suspend/hibernate on laptops isn't that great, but tbh I never had great results on windows either (macos is decent though).

And uptimes for desktop systems are similarly just limited by whenever there's a kernel update.

This is the opposite of what I've heard. Most often you hear of people installing Linux on old machines due to it performing better than Windows on low resources.

But not having to dual boot and just get both worlds at the same time definitely beats having to dual boot.

I don't think people are using WSL to avoid problems with dual booting. Dual-booting has become about as simple as it can be, thanks to UEFI, but it's still not exactly fun to have to close all of your open apps to switch to another OS to run just one app.

> You get much nicer window decorations if you use the wayland support instead of X11.

Wayland supports window managers ?

Sure we can speculate that 3% is not news. Again, it’s a one way conclusion: I concede if they want to axe a project deliberately, that could show up in the layoff, but projects won’t incidentally get impacted because of a 3%. The causal relationship would be the opposite.

If you mean stack ranking, the hard 20/70/10 bucketing was in force >15 years ago, but even then it didn't mean that those 10% automatically get fired.

Have you worked at any of these large companies? It’s really easy actually (practically, not emotionally). It’s usually very obvious and there’s consensus who the bottom 10% are. Politics would affect promotions much more than layoff.

But the knowledge about layoffs is at very high levels at the beginning

Managers learn about lay offs day or two before engineers

[flagged]

But most are just ignorant. Ignorance, not being a crime, of course.

Blunt and yet so true

Fortunately, Desktop users never need to recompile their kernel, it's really just a choice.

And I hope that you don't use WSL for servers :).

You can also use zsh in Linux ;)

You can’t? Just install UTM for a full VM one-click install (easier than wsl /install and two reboots) or any number of docker thingies that people build for the Mac.