Microsoft Open-Sources PowerShell & Brings It To Linux

[MartinN]

They should instead put bash on Windows by default and just call it a day.

My thoughts exactly )). For Microsoft to mean something to open source, they'd have to open source the NT kernel. Otherwise anything Nutella does is grasping at straws in the meantime.

[starshipeleven]

My thoughts exactly )). For Microsoft to mean something to open source, they'd have to open source the NT kernel. Otherwise anything Nutella does is grasping at straws in the meantime.

They can't. They rely on closed-source to hide bugs and backdoors.
If it is open they would actually have to work fixing that and removing backdoors, upsetting NSA.

[MartinN]

They can't. They rely on closed-source to hide bugs and backdoors.
If it is open they would actually have to work fixing that and removing backdoors, upsetting NSA.

Good point, didn't think of that .

[PsynoKhi0]

Yeah well... Still miffed by microsoft's license choice, specifically the lack of copyleft or any mechanism that would guarantee code modifications were contributed back. I could see subtly different yet widely incompatible versions down the road. Say, once education catches up and schools have been coerced into telling students that powershell is the One Shell To Rule Them All.
Let's use some caution here. That still comes from a company that made billions in close to 4 decades using vendor lock-in and triple E.

[starshipeleven]

Yeah well... Still miffed by microsoft's license choice,

They want to keep the same codebase with windows, as the whole point is running THE SAME shell that is multiplatform.
Also, they know full well that BSDs are a bunch of children and will never accept a non-permissive-licensed software on their pristine OS.

I could see subtly different yet widely incompatible versions down the road.

Self-defying effort, because see above. If it's not THE SAME as the one on Windows it's worthless crap. The point here is migrating stuff that was windows-only (scripts and all) to linux.
So the usebase is mostly coming from that.

Anyone already on linux is wildly unlikely to even give a fuck, because bash and python cover all usecases you might really want and are various orders of magnitude better in their own usecase.

Say, once education catches up and schools have been coerced into telling students that powershell is the One Shell To Rule Them All.

rampant paranoia.

[akincer]

Does that sound correct ?

No. Not even a little. You have no idea what my tools do or how much of the logic would be completely portable to another platform with minimal effort. Your post is full of ignorance.

There's no benefit to having Powershell on Linux.

Then you won't mind when others find benefit and make use of it.

[akincer]

Not only if you want to get super fancy. One of my biggest gripes with PowerShell (besides the bloat and ungodly startup times thanks to every application needing an object parser) is that even very simple tasks require you to memorize complex commands and their syntax.

Quoting myself from an earlier thread:

And in the meantime things got even worse. Now they are adding ssh support to PowerShell. It is a shell, not a crypto provider or a communications tool for crying out loud!


I think if you added dbus function to bash, you would arrive at something similar to PowerShell. Why does nobody do that? Because it would result in a mess. In the very rare case where one would need to talk dbus from a bash script, one would just call dbus-send etc. and keep all others free from that bloat.

If I want to pass objects around, I use Python.
If I want to do very simple tasks that at most require regular expressions on strings, I use bash.

If you want a single command to calculate an SHA sum in a minimalistic fashion, you can roll your own.

I think some people misunderstand what Powershell is. Despite having "shell" in the name, it's somewhere between a shell and a programming language. Newer versions have fixed many of the un-languagy parts of Powershell and moved the pendulum much closer to full fledged programming language, but older ones were certainly planted very firmly in the in-between.

I think some people forget WHY Microsoft essentially had to do it this way. Linux grew up on the CLI. Windows didn't. Bolting on enough functionality for Powershell to be anywhere in the ballpark as functional as what CLI junkies expect required them to throw in tons of stuff. At least that's my interpretation of it.

[ssokolow]

If you want a single command to calculate an SHA sum in a minimalistic fashion, you can roll your own.

I think some people misunderstand what Powershell is. Despite having "shell" in the name, it's somewhere between a shell and a programming language. Newer versions have fixed many of the un-languagy parts of Powershell and moved the pendulum much closer to full fledged programming language, but older ones were certainly planted very firmly in the in-between.

I think some people forget WHY Microsoft essentially had to do it this way. Linux grew up on the CLI. Windows didn't. Bolting on enough functionality for Powershell to be anywhere in the ballpark as functional as what CLI junkies expect required them to throw in tons of stuff. At least that's my interpretation of it.

Given some of the legacy decisions in how windows calls subprocesses, I can sort of understand that decision.

For example, yes, they could have made PowerShell a constellation of utilities in %PATH% by default, combined with a shell, but, unlike POSIXy OSes, the Windows APIs don't pass a pre-parsed argv array when calling the subprocess.

Instead, an un-parsed command-line string is received, then you just have to trust that the program is using the same "Microsoft's libc" parsing rules you expect. (There's a big cautionary notice about that in the Python subprocess module's docs)

[anda_skoa]

For example, yes, they could have made PowerShell a constellation of utilities in %PATH% by default, combined with a shell, but, unlike POSIXy OSes, the Windows APIs don't pass a pre-parsed argv array when calling the subprocess.

Instead, an un-parsed command-line string is received, then you just have to trust that the program is using the same "Microsoft's libc" parsing rules you expect. (There's a big cautionary notice about that in the Python subprocess module's docs)

Wow. I have been programming for decades now and never heard of that.
Which means the frameworks I've been working with on Windows did that for me.

Maybe .NET does that as well or do they use a different approach to overcome this for tools invoked by Powershell?

Cheers,
_

[ssokolow]

Wow. I have been programming for decades now and never heard of that.
Which means the frameworks I've been working with on Windows did that for me.

Maybe .NET does that as well or do they use a different approach to overcome this for tools invoked by Powershell?

Cheers,
_

Unless there are newer APIs I haven't heard of, they'd be relying on the fact that both .NET and PowerShell are built on the same libc (Microsoft's) to ensure the quoting grammars match.