Microsoft's Windows Subsystem For Linux Exits Beta

[mememe]

They've changed a lot since the days of Steve Ballmer and his crusade against the evils of open source. These days, their attitude seems to be that if they can make money off it, they'll take money off it.

So yes, it's a little embarrassing for them that a rival OS makes up such a large proportion of their cloud service. But if it's profitable, I'm sure they can live with a little embarrassment...

Well, Microsoft is a company and as such its primary motivation is to make money. Twenty years ago Microsoft was losing customers as Linux was making its ways into the mainframes and web servers of big companies. Since then the internet has drastically changed. Microsoft realized that there is more money to be made with cloud services and if you look at their recent financial reports you'll see that they are pretty damn successful with this new strategy.

What's probably more embarrassing is that more people are running the mobile Microsoft Office apps on Android than on Windows Phone.

[rabcor]

Microsoft confirms they still have no plans to support X/GUI Linux apps under WSL and you still aren't able to access Linux files from Windows, but could be improved upon in time. WSL is also not recommended for running production server workloads.

So it's essentially almost completely useless. No surprises here.

[Djhg2000]

I don't have any first hand experience with WSL, but a good friend of mine said he was able to use it together with Xming for running virt-manager and connect to his VM box. He even provided a screenshot and aside from some theme clashing it looked like it ran just fine.

[Delgarde]

There's generally two ways to look at the WSL...

The first and the most obvious one is that it's a pure Embrace, Extend, Extinguish play intended to allow Windows user to run Linux-only software and Linux developers to do their development on Windows. Knowing Microsoft's past with people like Steve Balmer calling Linux "Literally communism" when he was CEO of the company this would be the most likely reason if Microsoft was still being run by the company's old guard.

However when you remember what the current management has been much more open to open source and at the same had a strong focus on cost reduction, it spawns a completely different theory. This theory suggests that Microsoft is moving to using open more and more source software and open standards as a way to continue providing customers with not only the same products and services, but also at lower cost to themselves and maybe the customers as well.

I for one can't make a definitive choice between either theory as in my mind neither theory is either completely unlikely or very likely.

I'd say it's mostly the latter. If the company was still being run by the old guard, I'd say the former - but the company has evolved in directions that would have been inconceivable 10-15 years ago. Back then, their attitude towards Linux and open-source was an opposition that bordered on ideological... they saw it only as a threat that had to be destroyed before it could destroy them.

For a variety of reasons, the modern Microsoft is a more pragmatic creature. One part of that, I think, is that they've lost a few fights - both their web browser and their mobile OS have been soundly beaten in the market - and that's put them in a position where they've been forced into cooperation, unable to throw their weight around as much as before. And more recently, the rise of cloud has given them a new income stream - Linux might still be a threat to Windows, but Microsoft as a whole can make money off both of them, so they're not quite so reliant on Windows revenue as before.


I think that change is also mirrored in the open-source environment too... it's less ideology-driven than it used to be, less crusading, and more professionalism. Open-source is mainstream these days... whether it's server-side or web or mobile, few developers would think twice about using open-source libraries in their projects... and indeed, most will now be reluctant to touch anything closed. And in enterprise space, Linux servers are continuing to clean up the remains of the old Unix world... fewer and fewer clients are clinging to their HP-UX and AIX and Solaris systems, migrating to Intel + Linux as their old hardware expires.

[DrYak]

So it's essentially almost completely useless. No surprises here.

None of the regular Phoronix people are the target market for WSL.

WSL is done* so web developers can quickly locally test code that is targetted to run on the cloud (as currently, the cloud is dominated by linux-running virtual machines, even Microsoft's own Azure cloud runs Linux instances),
- but without needing to run said code inside a full-blown VM (e.g.: a complete Ubuntu install inside Virtual Box),
- or without needing to buy UNIX running hardware (Apple MacBooks, or laptops with Ubuntu/Fedora/etc pre-installed)
- instead they provide the strict bare minimum Linux kernel APIs as a "peronnality" to the Windows kernal to run linux ELFs un-modified.

So you're only going to have very high-level simple file I/O (forget about getting any Linux filesystem driver running. No BTRFS/EXT4/etc. for you), and some networking.
Basically enough to get Apache/NGINX/Node.JS/Python/etc. (miro-) servers running and connect to them with you regular vanilla windows Firefox/Chrome/Edge.

That could by coincidence be useable in a few other use cases : basically anything that is console-only and pure highlevel processing with no low-level access. E.g.: using scientific software compiled for Ubuntu to pre-process experimental data (only file I/O and console interface, nothing fancy).

So none of the linux/unix power users that dwell in this forum is going to be any interested in WSL.
Only the few web devs that are fed up with fireing up VirtualBOX just to test their latest React.JS web app on a real Ubuntu will be happy.

*: nowadays. Back in the beginning, WSL was supposed to be a layer to help microsoft port android run-time to Windows Phone, so they could get Android apps running on their Nokia phones. They didn't succeed, WSL is what they managed to salvage out of the effort.

I don't have any first hand experience with WSL, but a good friend of mine said he was able to use it together with Xming for running virt-manager and connect to his VM box.

In my limited experience (I was more interested in running scientific software than GUI), networking is more or less functional (at least if you don't require anything low-level or weird. Forget about running a multi-cast RTP WebTV).
So X-over-network to a win32 X server should be a viable solution to get GUI Apps running.

But again, if you're thinking about running GUI Apps, you're not WSL target audience anymore.
Better switch to some Unix OS (Ubuntu/Fedora/openSUSE-Tumbleweed or Mac OS X, depending on the laptop), or at least run a full Linux distro inside Virtual BOX.

One way to "get rid of something" is to find ways to make it slow and shitty so that it has a bad reputation. All Microsoft has to do is offer Microsoft Azure to as many as possible and then make it suck to make their Windows offering more appealing.

Ignoring all the other good argument against this idea (if Ubuntu VMs on Azure start to run slower than on AWS, most company will simply move their servers VMs elsewhere instead of rewritting everything from the ground up for windows), currently from what I've gathered WSL performance actually doesn't suck that much.

The biggest draw back of Windows (and thus any software layer that relies on the Win32 personality - like Cygwin) is that it completely sucks at multi-processing : creating and tearing down new processes / contexts takes a lot of time.
For WSL, Microsoft also introduced a new capability inside their kernel - pico threads - that their WSL personality can leverage to make multiple-procs more bearable.

I hate EEE and I hate Microsoft. But this particular move is harmless.

I would even add : EEE can't work for them now.

The EEE strategy needs a (almost-) monopoly situation to leverage.

It work back then to hold back java for quite some time : Microsoft had full control of the whole stack - not only the desktop computers, but the servers too.

The were able to make a Java "Etension", that could be available to devs in their Windows workstation, that could then be deployed to the enterprise's Micosoft IIS / WinNT based server, and used by Windows / Internet Explorer end-users.
It simply worked better for a potential victim of EEE if everything was kept within a microsoft "realm".

That has changed in the modern internet, where most of the servers are Linux VMs running on the cloud (mostly Amazon's), and most of the content is consumed on smartphones.
If microsoft decides to make some weird "extension" inside WSL compared to Linux, that extension will be completely useless outside of the Windows 10 running on the dev's workstation.
If microsoft's special flavour of Linux doesn't work anywhere else, people will simply move to what works every where else: keep the same server and clients, just change the dev env on the dev's laptop.

[smitty3268]

It boggles the mind how people can't see what's happening.