That said, no hate toward this project. Arch Linux is probably my favorite distro (although I'm on Xubuntu at the moment).
> Boot times are very short with SSDs so restarting is not a problem.
Personally - this is a question of individual preference, though - I do not mind the OS boot time itself very much (to a degree). Whether the system takes ten seconds or two minutes from reset/power-on to the login screen does not make much of a difference, to me, psychologically. But once I log in, I need to open all of these programs, make sure the windows are in the right place, SSH connections to certain machines are open, etc. I admit I am kind of obsessive-compulsive about this, but this is a far greater (psychological) barrier to rebooting than the OS boot time itself.
I do not use Windows at home, and my work laptop runs Windows 7 (I intend to keep it that way, too), so I have not been able to play with the Linux subsystem on Windows. But if you consider it as an alternative to Cygwin (which I do use), it sound kind of nice. Now if only Windows had a native, builtin X server... ;-)
EDIT: With Cygwin or the Linux subsystem, driver issues are not an issue, of course, because hardware is still managed by the Windows kernel and Windows drivers.
Hibernation should solve this issue, no? It certainly does on single-OS machines. I practically only shut down my desktop when I want to tweak the hardware and/or BIOS.
It's a little more annoying for dual-OS because there is a 'hibernate' command and a 'reboot' command, but I'm not aware of a 'hibernate-then-reboot' one, so you have to manually hit the power button after hibernating instead.
Now I can play around Tensor flow (in theory) and not have to worry about the crummy GPU support in nearly all hypervisors.
I've reached the point where I'd love to have Linux on my work machine, no Windows at all. That's only because for the first time Linux isn't this cool kids playground that I'm not invited to.
https://blogs.msdn.microsoft.com/vcblog/2015/11/18/announcin...
That's not necessarily doable for everyone, but for me that workflow is so much simpler than it would be with dual booting or a separate Linux VM.
Although I'm guessing a few devs are in the boat with "all company PCs must be windows." Gives them an out no?
I'm excited because everyone at my company doesn't develop on linux because "we know visual studio" and "setting up a build environment will take too much time".
With the linux subsystem for windows, we have the ability to push them to slowly start developing new systems on linux. So I'm super happy about it.
That's essentially how the windows subsystem works. It might not technically be a VM, but practically it is.
Would be cool if the Windows Linux subsystem can work like Parallels for OSX that can be blended into the general OS but also snapshots taken and if the user wants, bring the Linux environment in it's own window.
Consider full disk encryption on the GNU/Linux drive as well, since Windows can still access the disk.
(*Or a boot loader on a separate USB drive which is never plugged in while running Windows)
Moving to something like this one day makes me conflicted. On one hand I feel like I would betray open source, on the other hand I wouldn't have to restart my machine to play games...
I would really love if Qubes OS would support that, but it's not quite there yet.
Beyond native ports, one can use PlayOnLinux or just plain Wine with varying degrees of success.
Also, what are the advantages compared to solution like https://github.com/RoliSoft/WSL-Distribution-Switcher, which allows selecting different distributions, including Arch Linux?
Sorry to rain on your parade, but there is _no_ official Arch Linux image on Docker[1], as you can easily verify[2]. Use at your own risk whatever Arch Linux image you happen to find on Docker Hub, such as this one[3]. It might work or it might break, but it ain't official.
[1]: https://bbs.archlinux.org/viewtopic.php?id=214973
This is the first step: embrace. The next step: extend. Apparently it's already happening, as people scurry to adapt shit so that it works with WSL. Next you'll have companies requesting that all Linux software be "adapted" to work under WSL, and if it isn't, they won't buy it.
Once this happens, Microsoft can proceed to step 3: Extinguish. How? Easy - by adding incompatible shit to WSL. If companies succeed in forcing Linux software vendors to provide a WSL-specific version, they will have to be compatible with these WSL extension. Voila - now they are supporting a new Microsoft platform that Microsoft controls.
Back before many of you were born, Microsoft killed entire product categories and companies by providing free versions of Office when it was first introduced, and making it compatible with competitors' file formats. Is it free now? Hell no, it's their main source of revenue!
Can they repeat the same hat trick with Linux? Who knows, but they damn sure are gonna try. Anyone who thinks "Microsoft loves Linux" needs to take a history lesson. The only Linux Microsoft would love would have a dead penguin for a logo.
This also applies to Visual Studio Code.
But, I've tinkered recently with Windows 10 and WSL, and it's actually really neat. They've done a remarkable job making it work like a real Linux system in a lot of regards; my projects kinda bump up against the limitations of that (as they are administrative tools and perform actions as root and such), and so I can see areas where it's still incomplete, but I'm honestly shocked at how well it all works.
The only issue I had was with my Oculus Rift, but I suspect I could have fixed it by getting a dedicated USB PCI card to passthrough. Anyways, for AAA games it worked wonderfully with very near native speeds.
I love RHEL/CentOS dearly for servers, but needing to write a complex spec file and deal with the rpmbuild system just to get an RPM for a library is a pain. You end up with /usr/local full of stuff that's hard to update.
WSL looks like a nice replacement for cygwin, but that's about it for me.
