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

Install Win32 OpenSSH test release

github.com

Post reply on HN
Print view View original

31–40 of 47 posts

Re: Install Win32 OpenSSH test release

#31

Post by RayDonnelly »

mikemaccana wrote: post #20
YCOmega wrote: post #12

Microsoft is going about the whole SSH thing in a very backwards way. If all they did was make or back an official POSIX environment, all of this comes for free. Insisting that PowerShell be the focal point for all these changes is just bizarre. Just install msys2.

A half-bad Unix on top of Windows, with its own duplicate way of handling services, storage, users, permissions and everything else, is a pretty poor setup. Powershell (particularly the way it separates data from presentation) is one of the best shells on any OS. Being able to access it from Linux is a good thing.

As soon as Microsoft open sources PowerShell I'll be more than happy to take a look at it and perhaps add it to MSYS2. Until then for me and many like me, it's neither half bad nor half good, it lies orthogonal to the scale on which I judge software.

Re: Install Win32 OpenSSH test release

#32

Post by 0x0 »

mikemaccana wrote: post #22
quanticle wrote: post #9

Earlier quoted context omitted.

Powershell integration. Cygwin SSH drops you into a bash prompt running on the Windows box. Powershell is much more deeply integrated with Windows, and is more useful for actual system administration (e.g. configuring IIS, SQL Server, etc) than Cygwin's bash.

Asides from being an actual Windows app as you mentioned, it's also a current OpenSSH codebase.

But cygwin seems to have v7.1 too. Is it "just" a build that doesn't depend on cygwin.dll? (Although I can appreciate that's probably a nontrivial porting job). Genuinely curious about what's new here.

Re: Install Win32 OpenSSH test release

#33

Post by mortonpe »

This project is _not_ about bringing UNIX to Windows. It is about bringing the open ssh protocol to Windows. As a daily UNIX and Windows user in an enterprise environment, I can say that a fully integrated sshd is a welcomed addition. I have written and maintained many WinRM interop ruby gems and cannot wait until we can deprecate them in favor of a better and sane remote she'll implementation. This is a real milestone, keep up the great work.

As for Cygwin sshd, it does work if posix is the order of the day but if you need a reliable terminal (for things like powershell) that scales to hundreds of servers and many admins Cygwin simply fails the test. Clunky domain authentication, private key auth is a joke, powershell only works sometimes (thanks PTY), network tokens are non-existent, etc. the list goes on and on.

Re: Install Win32 OpenSSH test release

#34

Post by gfody »

YCOmega wrote: post #17
EvanAnderson wrote: post #14

Earlier quoted context omitted.

They purchased a reasonably nice POSIX environment when they bought Interix in 1999 ( https://en.wikipedia.org/wiki/Interix ). Interix was a lot of fun to use, and I really wish it still existed as a product. I used to have fanciful dreams of a POSIX-based Linux-like "distribution" of Windows that kept a lot of the things I like about the Windows NT environment (NTFS filesystem, kernel object manager, service control…

Yeah, it went through a pile of rebrands until Windows 8 when it ended up getting pulled in one of the preview releases IIRC. I've wanted similar to you just so that people could target some kinda baseline on Windows with FOSS. I mean, I get that coming from the other direction, powershell seems like a nice treat. And if you use it, colour me impressed. I know in some cases, it's the only option for doing script work…

I loved using SFU/Interix with Gentoo-prefix on XP and Win7 enterprise. I'm using Msys2 now and am pretty happy with it. There are mingw packages for practically everything but openssh is msys only. So this native openssh should give some speedup.

Re: Install Win32 OpenSSH test release

#35

Post by sveiss »

YCOmega wrote: post #12

Microsoft is going about the whole SSH thing in a very backwards way. If all they did was make or back an official POSIX environment, all of this comes for free. Insisting that PowerShell be the focal point for all these changes is just bizarre. Just install msys2.

Why is having PowerShell be the focal point for this bizarre? PowerShell is Windows' native command line interface.

Most of the scriptable administration tools (which are being heavily promoted as the primary way to administer Windows boxes these days) are PowerShell based, the tools for talking to Azure are PowerShell based, the new package manager in Windows 10 is PowerShell based, etcetera.

OpenSSH isn't being ported to Windows to make Windows overall more POSIXish; it's being ported to make managing Windows boxen from POSIX ones easier, and vice versa.

Re: Install Win32 OpenSSH test release

#36

Post by Sanddancer »

0x0 wrote: post #32
mikemaccana wrote: post #22

Earlier quoted context omitted.

Asides from being an actual Windows app as you mentioned, it's also a current OpenSSH codebase.

But cygwin seems to have v7.1 too. Is it "just" a build that doesn't depend on cygwin.dll? (Although I can appreciate that's probably a nontrivial porting job). Genuinely curious about what's new here.

It's an early alpha build. This is "just" getting to a point where they're working on a modern codebase. MS has said that they're going to add more functionality as they develop this into a much more useful admin tool.

Re: Install Win32 OpenSSH test release

#37

Post by brazzledazzle »

zrm wrote: post #23
EvanAnderson wrote: post #19

Earlier quoted context omitted.

You could do this right now w/ w/ a "Startup Script" or using Group Policy Preferences. The authorization to logon is, presumably, tied to the "User Right" to logon interactively and shouldn't have anything to do w/ the SSH server (since it's just userland).

Authorized keys in the context of ssh means the ability to logon without password using public key authentication. Now that I'm thinking about this, anybody know how they're doing public key authentication for domain accounts at all? If you don't use a password to logon, how are you supposed to get a Kerberos ticket to use domain network resources?

I think you can use "klist" with "get":

https://technet.microsoft.com/en-us/library/hh134826.aspx?f=...

Re: Install Win32 OpenSSH test release

#39

Post by jamiesonbecker »

zrm wrote: post #18

Feature request: Domain group policy setting for authorized keys. Make some ssh key(s) authorized to logon as some domain user(s) for every machine in the domain.

We're going to be porting our Linux/UNIX (python) shim [1] to PowerShell asap so that you can provide your users an actual UX for key management... integrating it with group policies would be really cool too. (However, it might be orthogonal to our goal of local admin accounts to eliminate dependence on centralized auth.)

1. Userify cloud ssh key mgmt - https://userify.com

Re: Install Win32 OpenSSH test release

#40

Post by Sanddancer »

YCOmega wrote: post #12

Microsoft is going about the whole SSH thing in a very backwards way. If all they did was make or back an official POSIX environment, all of this comes for free. Insisting that PowerShell be the focal point for all these changes is just bizarre. Just install msys2.

There's a huge difference in functionality and philosophy between powershell and posix. For one, everything in powershell is an object. What this means, is that when you pipe things together, you're not just piping output strings, you have iterators, arrays, closures, typed values, etc at your disposal. Additionally, in powershell, you can use arbitrary CLR libraries as if they were native. Posix shells, in contrast, tend to require you to write and build wrapper libraries to import the functions into your scripting environment.

So in short, using a posix shell would mean getting rid of all of the work MS has put into an insanely flexible command environment over the past several years. Insisting on posix just denies the flexibility to experiment and develop something which serves a different problem set, or approaches a problem from a new angle.

Post reply on HN