Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Windows Subsystem for Linux Architectural Overview

blogs.msdn.microsoft.com

Post reply on HN
Print view View original

31–40 of 43 posts

Re: Windows Subsystem for Linux Architectural Overview

#31

Post by viraptor »

poooogles wrote: post #30
viraptor wrote: post #27

Earlier quoted context omitted.

Does the fork performance really matter these days anymore? I mean comparing from one side to the speed of loading the app itself, fork is usually super cheap; on the other comparing to per-process async event handling, it's so slow pretty much nobody uses it this way for network apps anymore. What's the use case for a fast fork today?

Redis uses fork to save to disk.

Does fork on windows actually block parent for a long time, or does it only delay the child? I thought it was the latter, which shouldn't affect Redis. (but I may be wrong)

Re: Windows Subsystem for Linux Architectural Overview

#32

Post by mamon »

cududa wrote: post #25
venomsnake wrote: post #19

Earlier quoted context omitted.

Microsoft are insane about providing compatibility when it suits them. It will work - if they see a business case for it.

I was just thinking after reading this - sure there were some engineering marvels that enabled this, but for the implementation it was just rote work with the sheer manpower Microsoft can muster up. That in itself is just really impressive. Edit: giving it more thought, a Mozilla engineer recently demo'd an immediate mode rendering engine that him and one other dev wrote in 9 months, with feature parity of modern bro…

>> big engineering effort or a couple '10x developers' - pretty impressive

Or, being Mozilla engineer for quite some time he had enough domain knowledge to do that.

Or he found a clever way to reuse good parts of Mozilla code with just enough changes to avoid copyright infringement lawsuit.

Re: Windows Subsystem for Linux Architectural Overview

#33

Post by the_why_of_y »

cududa wrote: post #25
venomsnake wrote: post #19

Earlier quoted context omitted.

Microsoft are insane about providing compatibility when it suits them. It will work - if they see a business case for it.

I was just thinking after reading this - sure there were some engineering marvels that enabled this, but for the implementation it was just rote work with the sheer manpower Microsoft can muster up. That in itself is just really impressive. Edit: giving it more thought, a Mozilla engineer recently demo'd an immediate mode rendering engine that him and one other dev wrote in 9 months, with feature parity of modern bro…

Assuming you're talking about WebRender?

http://pcwalton.github.io/slides/webrender-talk-022016/

... which is notable because it is not immediate mode but the first retained mode renderer in a browser. Also, they used Rust to cut down on the time spent debugging threading issues.

Re: Windows Subsystem for Linux Architectural Overview

#34

Post by superobserver »

cremno wrote: post #13
superobserver wrote: post #7

Earlier quoted context omitted.

By delete, do you mean I can delete %APPDATA%\Local\Lxss from the Windows side? I don't think that will work, as I've already tried that.

Have you stopped lxssmanager? https://msdn.microsoft.com/en-us/commandline/wsl/faq#how-do-...

This might actually work. Thanks!

Re: Windows Subsystem for Linux Architectural Overview

#35

Post by novaleaf »

MichaelGG wrote: post #5

The Drawbridge[1] stuff is pretty interesting looking. I really wish MS would ship a simple sandboxing solution, so I can run arbitrary binaries and restrict them. The new Metro app stuff is cute, but doesn't help with the millions of existing binaries. Nor does it seem very user-friendly or useful, for that matter. 1: http://research.microsoft.com/en-us/projects/drawbridge/

it's not what you want, but i use Sandboxie to run untrusted stuff. it's a freemium product.

http://www.sandboxie.com/

Re: Windows Subsystem for Linux Architectural Overview

#36

Post by contextfree »

voltagex_ wrote: post #18
MichaelGG wrote: post #5

The Drawbridge[1] stuff is pretty interesting looking. I really wish MS would ship a simple sandboxing solution, so I can run arbitrary binaries and restrict them. The new Metro app stuff is cute, but doesn't help with the millions of existing binaries. Nor does it seem very user-friendly or useful, for that matter. 1: http://research.microsoft.com/en-us/projects/drawbridge/

A UWP sandboxing solution was demoed at Build this year and will ship as part of the Anniversary Update (SDK?). Apparently even Age of Empires II worked under it.

Centennial isn't really a sandbox in the sense I think the poster above you meant - while processes have their reads/writes to certain filesystem/registry locations virtualized by default, this isn't a security boundary as they ultimately run at medium trust and can do anything the user can do.

AppContainer is the security sandbox used by modern apps (aka Metro/UWP). It can be used independently of other aspects of the modern app model - e.g., Chrome uses it to sandbox content processes - although this isn't documented very well (which I guess was what GP was complaining about?) and it seems like trying to sandbox apps that weren't designed to be sandboxed, as GP was wanting, would have inherent compatibility problems?

Re: Windows Subsystem for Linux Architectural Overview

#37

Post by sievebrain »

Josteniok wrote: post #3
CurtHagenlocher wrote: post #2

It's interesting that this was enabled by the Drawbridge work. I was somewhat under the impression that Drawbridge was dead and/or superseded by other container-related projects. It's often weirdly hard to find out what's going on with some particular Microsoft project or technology, even for those of us who are employees. I was pretty excited about Drawbridge ~3 years ago when I was working with it but it subsequent…

I created an account just to reply to this. You said, "It's often weirdly hard to find out what's going on with some particular Microsoft project or technology..." I was just thinking that very thing today. Why is that?

Organising and propagating information throughout a large organisation takes significant effort.

Re: Windows Subsystem for Linux Architectural Overview

#38

Post by jibsen »

viraptor wrote: post #27
netheril96 wrote: post #14

So `fork` is still expensive?

Does the fork performance really matter these days anymore? I mean comparing from one side to the speed of loading the app itself, fork is usually super cheap; on the other comparing to per-process async event handling, it's so slow pretty much nobody uses it this way for network apps anymore. What's the use case for a fast fork today?

I don't know if it is due to various fork implementations being slow on Windows, but programs that spawn a lot of subprocesses like build tools can be really slow [1].

I would love to see some numbers for those from Bash on Windows.

[1]: https://gist.github.com/jibsen/7ebeddde3bc2bfd421b96ae53a824...

Re: Windows Subsystem for Linux Architectural Overview

#39

Post by cududa »

the_why_of_y wrote: post #33
cududa wrote: post #25

Earlier quoted context omitted.

I was just thinking after reading this - sure there were some engineering marvels that enabled this, but for the implementation it was just rote work with the sheer manpower Microsoft can muster up. That in itself is just really impressive. Edit: giving it more thought, a Mozilla engineer recently demo'd an immediate mode rendering engine that him and one other dev wrote in 9 months, with feature parity of modern bro…

Assuming you're talking about WebRender? http://pcwalton.github.io/slides/webrender-talk-022016/ ... which is notable because it is not immediate mode but the first retained mode renderer in a browser. Also, they used Rust to cut down on the time spent debugging threading issues.

Sorry - mixed up immediate/ retained
Post reply on HN