Halleluja! The Second Coming of Windows Subsystem For Linux blesses Insider faithful Microsoft has given its army of unpaid testers some Linux loving with the latest build of next year's Windows 10. Build 18917 arrived overnight with improvements to the narrator and options to stop the thing "optimising" the life out of a user's bandwidth during delivery. However, Windows Subsystem for Linux (WSL) 2 is the …
The thing that made WSL 1 attractive is that it *isn't* a virtual machine. It's a bona fide API translation layer that does a pretty good job at transparently accessing the host's filesystem, networking, and other service. If WSL 2 is just a virtual machine ... then it doesn't offer any more value than simply running an actual copy of Linux under Hyper-V or VirtualBox or VMware. We don't just want it to run on the same computer. We want it integrated. We want it to be so integrated that future versions of Windows would be able to simply run Linux software out of the box.
If WSL 2 is just a VM with a VHD ... there's no value anymore. I'll just uninstall it and run my own VM.
"We want it to be so integrated that future versions of Windows would be able to simply run Linux software out of the box."
Why? You can already run the Linux distro of your choice (including a home-built, from scratch, totally custom distro!) on the box of your choice. Why would you want to get Redmond that deep into the loop? Why give them that kind of control over your system(s)? What's in it for you? I'm not being snarky, those are serious questions ... I see absolutely zero use for this kludge, outside the marketards and bean counters at Redmond.
"and swapping between boxen/virtuals is a PITA"
What do you mean "switch". I run several VirtualBox VMs and ssh into them from 'screen' sessions on Cygwin. Which also gives me a decent (if a bit sluggish) Linux command line on Windows too. I can, of course, use WSL to ssh to the VMs but I have problems running 'screen' on it. Then "switching" is simply Ctrl-A+<window number>.
In any case, Cygwin is always the second thing I install on a fresh Windows (the first is Firefox which I then use to download everything else). I really don't understand the lack of love for Cygwin. It's not very efficient but it works very reliably. You can even use it to set up ssh server on your Windows box.
So the AC doesn't like a word, and so resorts to ad hominem. Why do I think it's real problem is that it can't answer my questions without collapsing the entire mental house of cards it has built up around WSL?
I'm sorry I made you think, Pobrecito. Hopefully it didn't hurt too much.
Why would you want to get Redmond that deep into the loop? Why give them that kind of control over your system(s)?
Because Microsoft.
That's why, it is what they do.
Halleluja! indeed ...
Dominance is the M$ path to everything and it is exactly where they are heading.
Has everyone forgotten everything that has happened in the last 30 years?
MS is slowly but steadily wrapping its slimy tentacles around Linux with this WSLinux, their acquisition of Github being another part of the scheme and Poettering a prime suspect in my book.
The writing has been on the wall forever and even so the usual TD&Hs are all very happy because now they can run Linux apps in Windows!
So cool !!!!
But whatever the holy fuck do you want that crap for !!! Eh?
Embrace, extend and extinguish may sound like a SciFi joke but it is not.
I'll say it again: this type of stuff does not bode well for the Linux ecosystem.
It will end up rotting it from the inside.
And by then the virus will be deeply ingrained inside Linux and far too late.
O.
"future versions of Windows would be able to simply run Linux software out of the box."
It's more easily done the OTHER way: a Microsoft-blessed version of WINE, running on a Linux kernel, with a WINE-like subsystem that implements all of Win32 at the API level, making native X11 and Win32 calls.
The rest of their ".Not" crap could then make Win32 API calls through the layer... and I bet it would STILL be faster than it is in Win-10-nic!!!
That's just too obvious, though. Remember MS did this with Windows 1.x, 2.x, 3.x running on top of MS-DOS. So now it would be Win-11 running on Linux! heh heh heh
Or you could just run .NET directly on linux or OSX.
"Remember MS did this with Windows 1.x, 2.x, 3.x running on top of MS-DOS"
This is the opposite of that, the linux sub-system is running on Windows in the same way that old versions of Windows used to run on top of DOS.
"It's more easily done the OTHER way: a Microsoft-blessed version of WINE, running on a Linux kernel, with a WINE-like subsystem that implements all of Win32 at the API level, making native X11 and Win32 calls."
It really isn't easier that way. The Windows kernel is fully modular and its much easier to run Linux thunking at kernel level under Windows that it is to run Windows thunking at kernel level under Linux. Running OS emulation in userland isn't ideal.
> "WSL 2 is aimed at encouraging developers to stick with the Windows 10 platform and..."
Putting a Windows (3.x) subsystem in OS/2 work so well for IBM.
At the time I was deciding on whether to develop for Presentation Manager or Windows. IBM's announcement made me decide on something that would run in both OSes.
WINE hasn't worked so well for developers wanting to so both OSes, but WSL may just do that.
OS/2 failed for different reasons. microsoft wrote the presentation manager, and OS/2 1.2 (releasing about a year before Windows 3.0) set THE standard for GUI environments.
Microsoft has broken that of course since Windows "Ape" and Win-10-nic and UWP and 2D FLATTY FLATSO McFLATFACE, but I digress...
The point is that OS/2 was actually SUPERIOR in every way, like a 16-bit NT in a lot of ways, a better API, and so on. But IBM marketed it for the PS/2 computer, no love for clones. THAT basically KILLED it, whereas Windows went for the clones AND the PS/2, and the rest became history.
Too bad MS isn't studying their past successes, though. THEN they would see that they're following IBM's ineffective tactics on this one.
I was inside the castle for pretty much the entire run of OS/2, from pre-1.0 to 3.0 Warp. From my perspective, what killed OS/2 was IBM's defiant adherence to the original design decisions, and a refusal to modernize. For example, IBM insisted it run on the 286, even though by the time it reached any measurable market, the 386 was ubiquitous. That let to numerous architectural restrictions the messed up the works. From a developer's viewpoint, the single input queue (from all Windows versions prior to NT) was a nightmare; this was the old cooperative multitasking model, where a single badly designed program could hang the entire OS.
Microsoft addressed all these issues in NT, but IBM refused to budge with OS/2. When Microsoft demonstrated they were serious about NT, everyone gave up on OS/2. It held onto that darned single input queue to the very end!
"The idea of running some form of Linux in Windows seems to keep coming up"
And the explanation given, including, it seems, from those who see themselves as the target audience, is that they're employed in situations where they have to run Linux to do their job but have to run Windows-only PCs. It may not be a rational solution but the irrationality isn't on the part of the users but their manglements.
I suppose from MS point of view it enables customers to continue resisting Linux boxes and discovering that they actually work very well and they could actually wean themselves of the MS teat. I can't imagine why they'd put the effort into it otherwise unless unless all that wishful thinking about making Windows a Linux core with a Windows API and interface on top.
In regard to the last I read elsewhere that this build also includes steps towards separating the Shell, presumably that awful-looking flat thing, from the rest of the OS, just like we've been doing with every Unix-based GUI for years. Strange times.
from being IN WINDOWS? Snappier than WHAT, MS-DOS 1.0 on a FLOPPY?
EVERY test I've EVER run that compares Windows to Linux or FreeBSD has demonstrated that Linux and FreeBSD file system performance is a *LOT* better than windows, by at least 10%. I haven't run those tests in a while, but a lot of numbers that came out at around the release of Win-10-nic showed that 7 was a tad faster than 10, and significantly faster than 8, and about the same as XP [which is what I ran the tests on, using equivalent hardware].
The biggest single problem with the windows file system SEEMS to be what I like to call "paranoid cacheing". Linux and FreeBSD will use ALL of the available RAM as a read/write cache if necessary, to limit the amount of actual I/O until it gets efficiently flushed to the disk, thus making the I/O faster overall. When you have to wait for a write to complete, it just slows EVERYTHING down.
And you see this a LOT with Windows. It's not hard to reproduce, not hard at all. I am not 100% sure that the problem _IS_ "paranoid cacheing" but everything I see tells me that Windows waits for physical write completes, and may even assume it CHANGED ON DISK and then would re-read it back again [instead of leaving it in a cache and relying on it NOT changing], whereas Linux and BSD do asynchronous write cache and generous read cacheing, 'lazily' flushing the cache to disk and journaling the file system to ensure file system integrity if the power goes out or something.
As a result, _I_ _CALL_ _B.S._ ON THAT CLAIM, Microsoft. Maybe WSL is "snappier" than CYGWIN, or Linux in a VM hosted on a windows box, but _NO_ _WAY_ is it "snappier" than LINUX ITSELF!!!
In fact, I think windows should run in a VM on Linux so it can get a FILE SYSTEM PERFORMANCE BOOST from the Ext file system. Similar with FreeBSD, hosting windows and NOT the other way around. UFS+J, ZFS file systems, WAY better than anything Micro-shaft can offer.
/me thought of YSL ties when I saw WSL. Dunno why.
The truth of the matter is hidden in the US export regulations. MS Windows has a Mass Market exemption for export. That places a few restrictions on it: It is not allowed to be real-time, it is not allowed to support C4I and it is not allowed to offer advanced networking features. The end result is that MS is forced to slightly cripple the whole thing - ditto with Apple Mac OS.
They have to do that, since they simply cannot process 1 billion export licences. Read the EAR, ITAR and MTCR for details.
The result is that open source operating systems - Linux and BSD - will always outperform Windows or MacOS.
"EVERY test I've EVER run that compares Windows to Linux or FreeBSD has demonstrated that Linux and FreeBSD file system performance is a *LOT* better than windows, by at least 10%. "
Well there are regular benchmarks out there of Windows 10 versus the latest Ubuntu and Windows 10 generally wins hands down for battery life, GPU and file system performance.
Citations please?
For graphics and battery life then you are probably right given the propitiatory nature of many hardware drivers and APIs needed to make them work to their best.
For file system and general performance it seems Linux is still a bit better than Windows 10:
https://www.phoronix.com/scan.php?page=article&item=windows10-okt-wsl
> The VM issue also rears its head with the file system which is now a VHD using the ext4 file system and configured with an initial size of 256GB.
Now I know Windows systems can get bloated but that's a tad large !
https://docs.microsoft.com/en-us/windows/wsl/wsl2-ux-changes
Ah - should say initial *max* size ...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
