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

Testing the Windows Subsystem for Linux

blogs.msdn.microsoft.com

Post reply on HN
Print view View original

121–130 of 142 posts

Re: Testing the Windows Subsystem for Linux

#121

Post by TazeTSchnitzel »

cryptarch wrote: post #102
skrause wrote: post #97

Earlier quoted context omitted.

So you were actually serious? I thought your first post was great satire. Poe's law strikes again...

I think the syscall translation part is an irrelevant hack, seems silly to act like that is what makes the project more useful because it does not guarantee performance, compatibility or stability.

The syscall translation makes it more useful than the previous UNIX subsystem, because it means it can near-perfectly emulate a popular POSIX-like system (Linux), and thus be compatible with most of its software, rather than being its own eccentric platform that existing POSIX software must first be ported to.

Re: Testing the Windows Subsystem for Linux

#123

Post by Someone1234 »

squarefoot wrote: post #79
hdhzy wrote: post #10

Earlier quoted context omitted.

Yes, but names starting with a trademark are a no-no. See [0] (bitcrazed works on WSL) and [1]. Also IANAL. [0]: https://news.ycombinator.com/item?id=13697918 [1]: https://www.andreas-jung.com/contents/dont-use-docker-in-git...

And sometimes even parts of a trademark can be risky. https://en.wikipedia.org/wiki/Linspire

Unclear what you think that proves. Microsoft paid $20m to buy their trademark and they changed their name.

Re: Testing the Windows Subsystem for Linux

#124

Post by Someone1234 »

yakk0 wrote: post #118
EpicEng wrote: post #25

Earlier quoted context omitted.

Yeah it's the 'for' in there that makes it wrong. It's not 'for' Linux, it's for Windows. I know what it is and I _still_ think of it backwards every time I read it.

it's the subsystem for running Linux. It's not actually Linux, and after enabling the subsystem you still have to actually get the Ubutntu image. That image is the Linux part.

We all understand how it works. Repeating that is unconstructive.

It is a bad name that is bad English. It was bad when it was "Windows Services for UNIX" too as an aside.

Re: Testing the Windows Subsystem for Linux

#125

Post by deadbunny »

adolfojp wrote: post #99
nrki wrote: post #81

Recently tried to use WSL to do some pair-coding tests in an interview, with screen & tmux. The damned terminal kept making characters disappear! I actually thought my interviewer was inadvertently deleting characters, or maybe trolling me. I had to switch to using an actual Linux system. MS have a ways to go before getting this right. I daresay I'll be on OSX by the time that happens though.

Did you try using WSL with the creator's update? The fixes were extensive.

You mean the update that hasn't even finished rolling out globally?

Re: Testing the Windows Subsystem for Linux

#126

Post by itomato »

ryenus wrote: post #87
ape4 wrote: post #2

I still have a hard time with bash-backwards name: "Windows Subsystem for Linux". Its a Linux subsystem for Windows.

"Windows Subsystem for Linux", though a confusing name, is still better than "Linux On Windows", because the later would lead to LOW, which is unfortunately very true, in terms of performance, they just don't want you to ever think in that way. Edit: seriously, I'm just kidding

Instead we have BOW

Re: Testing the Windows Subsystem for Linux

#127

Post by cryptarch »

skrebbel wrote: post #120
cryptarch wrote: post #101

Earlier quoted context omitted.

You can say that about Linux as well; the bread-and-butter of the CLI experience is still the GNU stuff. EDIT: Yes, there are non-GNU Linux distributions, but WSL is based off Ubuntu specifically and ships with a GNU userland by default.

I feel that you miss the point of WSL. Bash and coreutils have worked on Windows for years. In fact, they ship with git so most developers have had them for years. The point is compatibility much more than CLI UX. The ability to run that nodejs or Python project even though one of the 1352 subsubsubdependencies has a bug on Windows.

So it's supposed to be more stable than Cygwin? Is it that right now?

I thought Cygwin was very stable and included most things you could want, while also supporting Win7/8, which are the last versions I'm willing to run on my hardware.

Re: Testing the Windows Subsystem for Linux

#128

Post by cryptarch »

TazeTSchnitzel wrote: post #121
cryptarch wrote: post #102

Earlier quoted context omitted.

I think the syscall translation part is an irrelevant hack, seems silly to act like that is what makes the project more useful because it does not guarantee performance, compatibility or stability.

The syscall translation makes it more useful than the previous UNIX subsystem, because it means it can near-perfectly emulate a popular POSIX-like system (Linux), and thus be compatible with most of its software, rather than being its own eccentric platform that existing POSIX software must first be ported to.

So is it near-perfect yet?

I don't see the value if it's not more stable than "excentric" platforms like Cygwin and MSYS, but I guess I wouldn't use it either way because there's no way I'm installing Windows 10 on my hardware.

Re: Testing the Windows Subsystem for Linux

#129

Post by boylan »

dingo_bat wrote: post #67
SwellJoe wrote: post #53

Earlier quoted context omitted.

I used it for a while, when Fedora 24 refused to work reliably on my 4k laptop (it worked under 23, and now works again under 25). It was OK. I mean, it's an amazing feat of engineering, but there's a ton of little (and not so little) quirks that made it painful. Sharing files across the two systems was (maybe still is?) disastrous. I've heard of total Linux system corruption, but I personally had files disappear (th…

> There are good terminals Can you name some good ones? I am sick of putty/cygwin :(

Yeah, http://cmder.net/ - packages ConEmu and clink so your terminal is somewhat the same when in Windows or if you have a tab open with WSL.

Re: Testing the Windows Subsystem for Linux

#130

Post by codebook »

ulber wrote: post #66
codebook wrote: post #52

I gave up WSL after experiencing significant performance drop on GHC, Stackage. It was less than 1/10th of native for compiling Haskell code.

Here's the issue on GitHub if you'd like to track it: https://github.com/Microsoft/BashOnWindows/issues/1671 benhills describes the cause: For some context, I've looked at what causes this slowdown. For some reason stack has mapped an mind-bogglingly huge region of memory (I'm talking dozens of terabytes). When we fork we walk the entire address range to set up the new process's state. We have a design that should va…

Thanks for the info!. I will try again.
Post reply on HN