Windows Subsystem for Linux

[Rusky]

I interpreted it as "Linux detects that I have a touch screen, so it insists that I have to use it and doesn't give me the option to disable it".

Then you interpreted it wrong, specifically to further your completely baseless argument that Microsoft software is full of outstanding bugs while Linux is somehow not.

how easy it is to break it by importing conflicting values from a .reg file or network login

And this is a good thing, because if say I'm a desktop user and I just want to quickly share some files with someone over a network I can just right-click on them and choose "share"

More self-contradictory nonsense- when Microsoft lets you move settings around via .reg files or network logins, they're incompetent and evil; when Linux uses completely unstructured text files for config it's a pinnacle of software engineering.

You're still spewing vacuous garbage, except for the rare moments that you do say something specific but it's wrong.

[iansjack]

And this is a good thing, because if say I'm a desktop user and I just want to quickly share some files with someone over a network I can just right-click on them and choose "share" (or whatever the option's called in your file manager) and not worry about the underlying details, but if I'm running a file-sharing server I can configure things in greater depth to set the system up exactly as I want and need it to be.)

You're talking about Windows here I presume? That's indeed the way it works and, as you say, it's great. Another good aspect of Windows is the far finer degree of control over security that it allows compared to Unix-like systems.

Can you imagine, in Linux - by default - the administrator has unfettered access to users' private files. That degree of insecurity would never be acceptable in a business setting!

[Kazinsal]

File sharing in Windows is just as easy. Network shares are right click file -> properties -> sharing. Adding users to a file or folder's permissions is right click file -> share with -> specific people. Blanket permissions are dealt with in the same interface. Granular permissions are one more click away, and that's an equally straightforward interface that lets you check boxes to define filesystem ACLs, without ever having to mention "filesystem ACLs" and other administrative terms that'll just confuse the hell out of "normal users" who just want their colleagues on the domain to be able to make edits to their project documentation or whatever.

This is a far cry better than octal permissions, umasks, a pain in the arse FACL command line, and the inane crap that makes up SELinux.

[iansjack]

Security, and the ease of setting a very finely grained level of the same, is one of the major reasons that corporations choose Windows rather than Linux. In these days of data-protection laws and compliance requirements they need the best security mechanisms possible. This also goes for the ease of setting up a hierarchy of administrators, each having only the appropriate access to a particular set of objects.

[onlyonemac]

Then you interpreted it wrong, specifically to further your completely baseless argument that Microsoft software is full of outstanding bugs while Linux is somehow not.

As I said, either interpretation is equally likely without further information, but you chose to ignore that bit specifically to further your completely baseless argument that I blanketly discard anything with the word "Microsoft" in it.

More self-contradictory nonsense- when Microsoft lets you move settings around via .reg files or network logins, they're incompetent and evil; when Linux uses completely unstructured text files for config it's a pinnacle of software engineering.

In Linux I can process configuration files like any other text files, with utilities such as grep, diff, patch, and so on that make it easy to compare configurations and move them around. Linux applications also tend to respond better to errors or discrepancies in their configuration files, and the configuration files are almost always structured such that discrepancies cannot in fact occur. On the other hand, the Windows registry is very prone to corruption by importing an erroneous .reg file (I know - I've broken three Windows systems that way).

[onlyonemac]

Granular permissions are one more click away, and that's an equally straightforward interface that lets you check boxes to define filesystem ACLs, without ever having to mention "filesystem ACLs" and other administrative terms that'll just confuse the hell out of "normal users" who just want their colleagues on the domain to be able to make edits to their project documentation or whatever.

Only if you use Windows Professional, AFAIK. Also, the Windows/NTFS ACLs are so complicated that even when I set all of the settings exactly the way I want them, they just get overridden by "inherited" settings from parent files all the way up to the root of the filesystem. And I also can't seem to control who is able to change the ACL.

[zdz]

Then you interpreted it wrong, specifically to further your completely baseless argument that Microsoft software is full of outstanding bugs while Linux is somehow not.

As I said, either interpretation is equally likely without further information, but you chose to ignore that bit specifically to further your completely baseless argument that I blanketly discard anything with the word "Microsoft" in it.

More self-contradictory nonsense- when Microsoft lets you move settings around via .reg files or network logins, they're incompetent and evil; when Linux uses completely unstructured text files for config it's a pinnacle of software engineering.

In Linux I can process configuration files like any other text files, with utilities such as grep, diff, patch, and so on that make it easy to compare configurations and move them around. Linux applications also tend to respond better to errors or discrepancies in their configuration files, and the configuration files are almost always structured such that discrepancies cannot in fact occur. On the other hand, the Windows registry is very prone to corruption by importing an erroneous .reg file (I know - I've broken three Windows systems that way).

You're kidding right? I can't remember the exact scenario, but I once started to learn ocaml so I went on and tried the beginner install tutorial from realworldocaml. And somewhere in the middle of the installation process something went wrong and utop refused to install. I remembered searching for that problem and I found it pretty quickly on one of the official communication channels marked by the devs as "not really a problem". Okay... I don't remembered what it was, but after more hours lost on google trying different things I figured out that I had *something* (again, can't remember what) at a version newer than that expected by utop. Nowhere in the error it gave me was that mentioned. And that wasn't the only time I lost hours trying to make something work. So yes, Linux applications tend to respond better to dicrepancies in the configuration files as long as the environmnet they are installed on is that perfect scenario some dev had in mind.
I'm pretty good with Ruby, but installing RVM is always a adventure for me because I just can feel that *something* will happen and nobody would be able to tell exactly why. I have somewhere a file with things to do if X application dies during install / update. Because that's how easy it is to make something work.
Let's not talk about convinging Ubuntu to see the shared folder did in vmware. Sometimes it works, sometimes it doesn't. And there are a handfull of solution for this problem. Each working if the planets are alligned. I can go on and on with this.
On the other hand, because I'm not so lost in my own believes: there are times when something crashes on Windows with the best error message in the world: "Something happened". But historically speaking, I'm able to fix those quicker than a Linux error report that covers 300 lines and tells me nothing Useful in the end.

Only if you use Windows Professional, AFAIK. Also, the Windows/NTFS ACLs are so complicated that even when I set all of the settings exactly the way I want them, they just get overridden by "inherited" settings from parent files all the way up to the root of the filesystem. And I also can't seem to control who is able to change the ACL.

That is actually documented. But when Microsoft expects from you to read the damn manual, they are evil...

[iansjack]

Only if you use Windows Professional, AFAIK.

Which demonstrates that you don't know. Now there's nothing wrong in that by itself, but when you criticize Windows so roundly from a position of profound ignorance then there is something wrong.

Also, the Windows/NTFS ACLs are so complicated that even when I set all of the settings exactly the way I want them, they just get overridden by "inherited" settings from parent files all the way up to the root of the filesystem. And I also can't seem to control who is able to change the ACL.

Once again, you are blaming Microsoft for your technical incompetence. That's really not fair. And, as already mentioned, Microsoft documents these things in exquisite detail. Documentation is another area where Windows is far superior to the mish-mash of information provided for Linux.

[glauxosdever]

Hi,


Please, stop fighting already! You are all of you aware that every operating system has its flaws. Don't try to desperately prove wrong the ones that disagree with you, they have used the operating system in question and they have some experience. Take the things criticised here by those that disagree with you as an opportunity not to repeat them in your operating systems (you are OS developers, aren't you?).

Not liking something because of philosophical reasons is fine. Trying to desperately pose technical reasons that make something look inferior even if it is better in that specific field is not fine and just makes you look silly.


Regards,
glauxosdever

[onlyonemac]

I can't remember the exact scenario, but I once started to learn ocaml so I went on and tried the beginner install tutorial from realworldocaml. And somewhere in the middle of the installation process something went wrong and utop refused to install. I remembered searching for that problem and I found it pretty quickly on one of the official communication channels marked by the devs as "not really a problem". Okay... I don't remembered what it was, but after more hours lost on google trying different things I figured out that I had *something* (again, can't remember what) at a version newer than that expected by utop. Nowhere in the error it gave me was that mentioned. And that wasn't the only time I lost hours trying to make something work. So yes, Linux applications tend to respond better to dicrepancies in the configuration files as long as the environmnet they are installed on is that perfect scenario some dev had in mind.

It's not the developers' problem if your distro's package maintainers don't specify the correct version of the package's dependencies when they prepare the package, or if your package manager doesn't install the correct version of the dependencies.

Let's not talk about convinging Ubuntu to see the shared folder did in vmware. Sometimes it works, sometimes it doesn't. And there are a handfull of solution for this problem. Each working if the planets are alligned. I can go on and on with this.

Don't use a non-standard feature of your VM host that requires installation of extra software and runs over a "backdoor" connection between the host and the guest if you want things to work reliably. Rather use a standard file-sharing protocol, such as Samba or NFS.

there are times when something crashes on Windows with the best error message in the world: "Something happened".

At least my computer doesn't say silly things like "something happened" when it crashes, but actually gives me enough information to phrase a Google search that's going to lead me to more specific answers than "make sure that all the latest updates are installed, run a malware scan, and if that fails then check for driver conflicts".

That is actually documented. But when Microsoft expects from you to read the damn manual, they are evil...

Perhaps they shouldn't make it so difficult to find and understand the documentation. In Linux I can just type "man <name of configuration file that I'm editing>" and usually I'll get the documentation that I need, and if not then "man <name of application to which configuration file belongs>" always does the trick.

[onlyonemac]

Only if you use Windows Professional, AFAIK.

Which demonstrates that you don't know. Now there's nothing wrong in that by itself, but when you criticize Windows so roundly from a position of profound ignorance then there is something wrong.

Let me tell you what I do know:

Back when XP was still a current Windows version, we had a Windows XP Home Edition computer (NTFS filesystem), and there was no option to configure anything like ACLs but only to "make this folder private" (which made the computer crash whenever I tried to use it, because there were too many files and subfolders inside the folder). A few years ago, my mother got a Windows 7 Professional computer (also NTFS filesystem), and there was the option to configure the ACLs (that I can never manage to configure correctly). This means one of two things:

• Windows XP didn't have ACLs but Windows 7 does
• Windows Home Edition doesn't have ACLs but Windows Professional does

Either that means that I have to pay more for a basic security feature, or Microsoft were severely behind with implementing a feature that Linux has had for a long time.

[onlyonemac]

Also, get this:

About a year and a half ago a friend came to visit. At the time, he dual-booted his laptop with Windows 8 (or Windows 8.1, can't remember which but it doesn't matter for the point of the story) and Ubuntu 14.04. When he was here, I asked if I could try out Windows 8 on his laptop as I had not used Windows 8 before. He had not booted Windows for about three months, and last time he had used it it had been working fine (as well as Windows ever does). As soon as he logged in, the mouse pointer froze, then a few second later the screen went blank, and then it BSODed. He rebooted it and it hung on whatever animation it does when it's booting. I don't know what it was doing that caused it to crash (it wasn't even connected to the internet), but it was obviously related to how long it had been since the system had been used last, which has never been an issue with any of my Linux systems even though some of them have at times gone almost a year without being booted.

Oh, and a few weeks later he told me about how he had tried to migrate another Windows system from a physical computer to a VirtualBox VM, which failed due to "driver issues", something which has again never been a problem for my Linux systems.

[zdz]

Hi,


Please, stop fighting already! You are all of you aware that every operating system has its flaws. Don't try to desperately prove wrong the ones that disagree with you, they have used the operating system in question and they have some experience. Take the things criticised here by those that disagree with you as an opportunity not to repeat them in your operating systems (you are OS developers, aren't you?).

Not liking something because of philosophical reasons is fine. Trying to desperately pose technical reasons that make something look inferior even if it is better in that specific field is not fine and just makes you look silly.


Regards,
glauxosdever

You're right But I'm a bit sick of the way most people with this passion look at Windows and related products, when most of the time they only have this idea of "Windows bad, Linux good" in mind without even knowing the details. I use both, because I need both. Not that I couldn't do everything on one OS, but because I prefer to use the best OS for the tools I need.
For me it doesn't matter if it is Linux on Windows or Windows on Linux, I appreciate it. It's a bit of a shame that this discussion got so far away from the initial theme.

It's not the developers' problem if your distro's package maintainers don't specify the correct version of the package's dependencies when they prepare the package, or if your package manager doesn't install the correct version of the dependencies.

That storry was a response to the "easier to configure, easier to maintain, less chances of something to go wrong", but you choose to ignore that. The developer should offer me clear messages when something fails. I have to mention that I needed the newer version of whatever, and I wasn't able to have two different version at the same time. But it's good that I have a lot of config files in a lot of (undocumented) places. Yey!

Don't use a non-standard feature of your VM host that requires installation of extra software and runs over a "backdoor" connection between the host and the guest if you want things to work reliably. Rather use a standard file-sharing protocol, such as Samba or NFS.

But I don't need SAMBA or NFS. I want to point vmware to a .ISO, let it install it and do it's magic. For Windows guests this is simply resolved by having the shared folder appear as a network folder.

Perhaps they shouldn't make it so difficult to find and understand the documentation. In Linux I can just type "man <name of configuration file that I'm editing>" and usually I'll get the documentation that I need, and if not then "man <name of application to which configuration file belongs>" always does the trick.

Because reading a full manual it's easier from the command line than from a text / pdf / html etc file. Go ahead, read the Intel manuals in this manner. We are not talking about "let me see what this command is doing", but "I need to learn how to use this complex feature of the OS".

Also, get this:

About a year and a half ago a friend came to visit. At the time, he dual-booted his laptop with Windows 8 (or Windows 8.1, can't remember which but it doesn't matter for the point of the story) and Ubuntu 14.04. When he was here, I asked if I could try out Windows 8 on his laptop as I had not used Windows 8 before. He had not booted Windows for about three months, and last time he had used it it had been working fine (as well as Windows ever does). As soon as he logged in, the mouse pointer froze, then a few second later the screen went blank, and then it BSODed. He rebooted it and it hung on whatever animation it does when it's booting. I don't know what it was doing that caused it to crash (it wasn't even connected to the internet), but it was obviously related to how long it had been since the system had been used last, which has never been an issue with any of my Linux systems even though some of them have at times gone almost a year without being booted.

Oh, and a few weeks later he told me about how he had tried to migrate another Windows system from a physical computer to a VirtualBox VM, which failed due to "driver issues", something which has again never been a problem for my Linux systems.

You have to understand that most BSODs are caused by faulty drivers, not by the OS. Have you had the curiosity to look at the memory dump to pinpoint the actual problem? I've never experienced BSODs caused directly by a Windows driver. Also, long boot times, weird behaviour, unexpected BSODs, there might have been some malware around. I had an old laptop (i5, 4G RAM, can't remember anything else) which booted Windows 8.1 in 15-20 seconds. This tells nothing.
We ran a lot of different tests on our drivers and there are crashes that reproduce in such weird condition that you can only ask yourself if we can catch them all. Sometimes it's not even the fault of one driver, but some weird interaction between two or more. You can't summarize it to "I had a friend who had a BSOD".

[iansjack]

Let me tell you what I do know:

Agan, I'm afraid, you are just continuing to demonstrate a complete lack of knowledge about Windows. NTFS has had a full set of fine-grained access permissions at least since Windows 2000 and no doubt before that.

I'd be only too pleased for Linux to catch up with Windows in this respect but it's nowhere near it yet.

[Rusky]

I blanketly discard anything with the word "Microsoft" in it.

Alright then, what has Microsoft done that you don't consider evil?