It happens to be the shell that coincides with Microsoft making an actually-somewhat-serious attempt to make the CLI a first class administrative interface. CMD.exe certainly sucked; but what really ruined it was how often you would run into the problem that there simply wasn’t a command line way to do something; or if you were really lucky there would be a slightly eccentric utility stashed in the VALUEADD directory of an old Windows install disk or included in an admin toolkit somewhere. With Powershell, Microsoft has made a real effort to ensure that if something needs administrating; there exists a knob in powershell to administrate it.
Architecturally, it’s a shell built to be object oriented, rather than good at string mangling. There are some mechanisms for mangling strings; but what gets piped around and emitted and ingested are actually objects. From my perspective as a non programmer who just needs to bodge a script together sometimes, this is a mixed blessing: passing strings around occasionally requires really annoying caution(could the result of that command you ran possibly include characters that need to be escaped so they don’t munge everything else?); but it is conceptually pretty trivial. OO is harder to get my head around.
I haven’t really done anything with it yet – but I’ve hit all the pre-reqs and fiddly bits in the configuration to get it installed and restarted and installed and restarted and running enough to find the whereis command.
I’ve got some brief notes on it, that I hope to expand.
My goal? Run some NodeJS image libraries that require Linux packages that I’ve never gotten to work on Windows.
I think the biggest win for PowerShell is it’s essentially a front end to .NET. That’s incredibly powerful right there. You can directly address .NET objects and types. You can emit and compile C# code on the fly (great for getting direct access to Win32 APIs). There’s many other great things you can do.
For instance, want a GUID on the fly?
Done!
Need to convert a long to an IP address (something I needed to do today):
[ipaddress]::new(16777343)
Done!
It’s well supported and constantly being improved.
I admit that my perspective on it is very much that of an administrative script-bodger, so most of the elegance is lost on me; but the thoroughness with which the bits of of Windows that need twiddling have been exposed in Powershell for twiddling is highly appreciated.
On a theoretical basis, I can appreciate the virtues of the fact that MS has gone from having pretty much the lousiest scripting environment available to having a shell heavily anchored in its (mostly well received) software development tools; but I’m too much of a lightweight for that to really change how I use it very much. This hardly means that “your shell scripts can seamlessly range from basic batch files up to basically full .NET development” is a bad thing; it’s just not a thing that I’m enough of a developer to make any use of.
The fact that Microsoft is taking a fairly serious stab this time around at making “If you can do it in the GUI, you can script it, period; plus you can script a whole bunch of other stuff.” actually true, by contrast, is something that has saved me considerable weeping and gnashing of teeth. If I sucked less at programming the elegance of the implementation would probably matter more to me; but since I don’t, the big change from my perspective is that MS is actually attempting to make CLI administration feature complete; which has historically not been the case, at all, on Windows.
(TL;DR: I in no way deny praise of Powershell’s elegance/architectural power/etc.; but from the perspective of someone with a limited ability to appreciate those virtues, the big deal is that Powershell coincides with MS moving to make CLI a first class UI and expose a lot more of what traditionally had no CLI mechanisms to work with.)
Last weekend for an emergency job I used AWK for the first time in like 20 years. It is astonishing how much you can forget in 20 years. (I’d also forgotten how limited AWK is. As much as I used to appreciate how lean *nix utilities are when working with 64k of RAM, that probably isn’t necessary any more.)
I’m a BSD head. Take that in to consideration when I make this statement.
Fuck you linux. Get your userspace in sync with the kernel. Don’t just pile a bunch of toys on top of it and call it an “operating system”. (Problem being “your” is distributed through a bunch of random enthusiasts.)
I’ve been running LinuxMint 17 for… A long time now, and suddenly on a restart TeamViewer stops working (yes, I know I should do something better, like SSH or what have you, shaddup). And also most software seems to be unable to install.
As far as I can tell, all my sources have gotten corrupted, along with my GStreamer framework, which means Firefox won’t even run HTML5 video in the browser correctly anymore.
It’s starting to look like I need to backup and reinstall. Which is something that I’m only used to doing under Windows. Typically, if I don’t monkey around with shit too hard, linux doesn’t get broken this way for me. I don’t even know how this could happen all at once.
I’m sympathetic to your point, but in this case it actually wasn’t a bunch of random enthusiasts, it was Red Hat - the supposedly standard reliable expensive distro - that broke linux POSIX compliance in their PAM/NSS solution. Ever since they created Fedora they’ve been getting less and less in touch with the server room customers that made their success… it seems the Fedora devs aren’t much different from random enthusiasts in any positive way.
