Can you make WSL interface with windows installs? [duplicate] - windows-subsystem-for-linux

How do I avoid installing same programming languages both in WSL and Windows10?
I am thinking about using WSL as a dev workspace. However, I realized I will need to install Node.js, Python, create-react-app, and so on in WSL even though my windows 10 already have them installed.
It would be helpful if you could spare me some advice.
Thanks.

To some degree, it depends on what type of development you are doing. Given your example languages/tools, I'm going to assume that most of your development is platform agnostic, web-development, etc.
My recommendation is to go all-in on WSL and install the Linux versions of the tools you use (with some notable exceptions covered below).
Uninstalling the Windows versions is recommended, but not strictly necessary. I recommend uninstalling because I continue to see a number of questions across the Stack sites where it becomes apparent that the Windows version of Node or Python is getting called from inside WSL. It's likely that some tool, such as nvm or equivalent, attempted to prepend the Windows Node or Python location to the Linux path.
This causes problems, as the Windows versions Node and Python understand Windows paths and processes. When you call them from the Linux shell in WSL, the shell/OS uses, of course, the Linux versions. And Windows Python just won't understand something like /mnt/c/Projects. It needs C:\Projects. You can work around this with utilities such as wslpath (automatically installed in some WSL distributions, installable in all others), or you could just manually adjust the path. But ... why go through the hassle if you don't need to.
Just use the Linux versions, with the corresponding Linux paths and instructions. Most development tools, tutorials, instructions, etc. are going to "default" to the Linux doc. It will typically be more complete, more up-to-date, etc.
And, of course, the Linux command-line experience is (subjectively, sure) far-and-above better than PowerShell. Don't get me wrong, I like PowerShell, but I like PowerShell even better when I call it through WSL (powershell.exe or pwsh.exe), since I can take advantage of Linux niceties like less (or bat), jq, and many others.
Not to say there aren't WSL caveats that you have to get used to. Be prepared to run into a few snags here and there (lack of Systemd support, permissions, filesystems, inotify), but most everything has a workaround that you'll typically find here on Stack (Stack Overflow, Ask Ubuntu, Unix & Linux, and/or Super User) if you search.
And for those "notable exceptions" I mentioned, I recommend installing:
Windows Terminal (available in the Microsoft Store), which will provide an upgraded terminal experience for WSL.
The Windows version of Visual Studio Code -- I've seen a question from someone here who tried to install the Linux version. It's just not necessary. Microsoft has done a great job of integrating the Windows version of VSCode with WSL. Just install the "Remote Development" extension pack, which includes the "Remote - WSL" extension.

Related

How to turn on inotify in WSL2?

inotify worked in WSL1. Then it was deliberately turned off in WSL2 due to an unsupported feature in a GNU software which is now solved.
How can inotify be enabled or turned on in WSL2?
inotify is supported in WSL2, but only on the Linux-based ext4 filesystem. Where you may be having trouble is that it does not work on Windows drives which are mounted using the 9P protocol (e.g. /mnt/c) or symlinks to any files on those drives.
I'm unaware of this being (per the question) due to "an unsupported feature in a GNU software which is now solved", nor it being "deliberately turned off". It's my understanding that the WSL team just hasn't "plumbed" (their word from the 2019 Build Conference) it in 9P.
It does, as you mention work in WSL1 on Windows drives mounted via drvfs, and using WSL1 is still a viable option for many development tasks. Of course, this is only necessary if you require that your watched files be on a Windows drive. Also note that WSL1 really used the Windows drive for both the Linux filesystem (via an overlay of sorts in your WSL1 directory), so if inotify worked for one, it likely worked for both for the same reason (same implementation of syscall translation).
The simplest solution, though, if it meets your workflow, is to just move your project somewhere on the WSL/Linux/ext4 filesystem, such as under your $HOME folder (again, not using a symlink).
As for how to enable it, I don't think that's going to be possible. While the 9P client is open-source and included in the WSL2 kernel Github project here, the server that runs in Windows and provides access to those drives is, as far as I know, still closed-source.
For more details, see this answer.

Wsl2 why develop in this environment

I understand the principal of having Linux on your windows box over using a VM.
I have seen articles on using vscode in wsl with the vscode wsl extension.
My question is why would I do this over just using vscode on Windows.
Might sound like a silly question and I hear people saying I can now develop on Linux where my company is is windows, I just don't understand what the benefits are
It all depends on what you are developing, if its C# with the target operating system being windows or azure there is little benefit.
Personally I work in nodejs, our target deployment OS is centos running in AWS-EC2, so it makes total sense. There is a lot of tooling that runs better in a linux environment than in a dos/powershell environment. Also when testing we are running the code in an environment that is close to our target env which reduces the chances of unforeseen issues further.
Does this make sense? what are you developing day to day and what is your target environment?

What is the best way to remotely edit a file using VS code?

Currently, I have two machines, one with Ubuntu in the company and one with Mac OS at home. Sometimes I would like to work at home while accessing the Ubuntu machine in the company. I can ssh into the Ubuntu machine and navigate and compile there. However, when I actually want to edit some cpp source codes, I realize that the editor (VS code) is actually opened in the Ubuntu machine, so I cannot view it from Mac. What should I do if want to edit files remotely on my Mac through VS code?
Though many of the answers mention using version control tools like git, it can be hard to use in my specific case. The problem is that the building environment of my company is Linux, so most of the building tools I have can only run on Linux. This means that I can only compile my source codes in Linux. If I use git, then every time I want to compile and debug my codes, I have to commit and push with my Mac, and then pull and test on Linux. This can be time consuming if want to incrementally modify, test and debug my codes.
Use some version control system like git. Then you might edit and compile at home (provided your code is portable between Linux & MacOSX, e.g. because it is POSIX compliant).
You could install some X11 server on your Mac and use ssh -X to access the remote Ubuntu machine (then run a GUI or editor remotely, e.g. ssh -X remotelinuxhost.company.com emacs). However, that requires good bandwidth and latency between your home computer and the remote one.
BTW, you might use some other source code editor, like emacs (it is capable of remote editing) or vim.
Since Linux and MacOSX are both POSIX systems, it is usually (but not always) easy to port source code from Linux to MacOSX and write source code compilable on both systems. BTW, many Linux frameworks (e.g. Qt, GTK, POCO, Boost, etc...) and build systems are usable and ported to MacOSX. Some Linux system calls (listed in syscalls(2)) are not available on MacOSX (e.g. signalfd(2)...)
Of course you could install Linux (perhaps inside some VM) on your Apple laptop.

How to install recent mono and monodevelop?

I tried to install mono and monodevelop on centOS 6.3.
After many hours I was able to install mono but failed with monodevelop.
I'm really astonished how difficult and time consuming it is, to get a recent mono/monodevelop version on linux installed.
Is there nobody willing to write and maintain an install/compile tutorial to get the most recent mono/monodevelop/monodata/ASP.NET MVC/... version on the major linux distributions (Centos, Ubuntu, Suse, Debian) installed?
I think many people developing on Windows (with limited linux knowledge) would like to start using mono, if the boarding hurdle would be somehow lower.
It may be the most important to make Mono more used and more visible.
Please, write a tested tutorial (script) for compiling mono/monodevelop.
Thank you!
I have created a project on Open Build Service, which produces builds of the latest MonoDevelop 4.0.10 for Debian, Ubuntu, CentOS, and Fedora.
see https://build.opensuse.org/project/show/home:tpokorra:mono
For installation instructions with apt-get or yum, see:
http://software.opensuse.org/download/package?project=home:tpokorra:mono&package=monodevelop-opt
I hope this will increase the usage of MonoDevelop on Linux Desktop environments.
Monodevelop 4.
If you use any *buntu. Check this.
"You can open up the terminal and install it via the following:
1. sudo add-apt-repository ppa:keks9n/monodevelop-latest
2. sudo apt-get update
3. sudo apt-get install monodevelop-latest"
http://mono-d.alexanderbothe.com/?p=101
Xamarin should be doing a better job at publishing the linux packages in a one-click manner. I don't care what linux distro (SuSE, RHEL, CentOS, Ubuntu etc) - just pick any one as the supported one and publish for it. It seemed that it used to be SuSE but even that has old packages as seen within Zypper/YaST.
Update Mono framework
Having said that, to update the Mono framework itself, without letting go of the package managers try this. This will work as long as the project dutifully publishes the RPMs. You don't want to build from source since it's a more fickle process and the setup distracts from your real objective (i.e. develop).
Obviously, please replace the URL below to what will be latest by the time you're reading this.
mkdir mono-rpms
cd mono-rpms
wget --reject "index.html*" -nd -r -e robots=off --no-parent http://download.mono-project.com/archive/3.2.3/linux/x64/
sudo zypper install *rpm
Update MonoDevelop (the IDE)
Timotheus Pokorra's answer indicates he's filling in some of the usability void left by Xamarin (Thanks Timotheus!!). You can install MonoDevelop via
http://software.opensuse.org/download/package?project=home:tpokorra:mono&package=monodevelop-opt
Note that on SuSE I get the error
Problem: nothing provides liberation-mono-fonts needed by mono-libgdiplus-opt-3.0.12-7.1.x86_64
Solution 1: do not install monodevelop-opt-4.0.12-5.2.x86_64
Solution 2: break mono-libgdiplus-opt-3.0.12-7.1.x86_64 by ignoring some of its dependencies
I (very reluctantly) selected to break the dependency. Note that I already had liberation-fonts (via sudo zypper install liberation-fonts). I don't know if its the same/different as liberation-mono-fonts. Anyway, hope Timotheus fixes it when he has a moment.
I'm not sure if you've already seen this, but this may help:
http://www.mono-project.com/Parallel_Mono_Environments
The most common problem that new developers have when coming to Linux from systems like Windows is not properly setting up their environment variables and so when they do the standard ./configure && make && make install routine, when it involves a number of source packages (like Mono does), any package that depends on the core package won't pick up the correct location for that base package.
Your question really doesn't explain what parts you found confusing or difficult so it's hard to address those issues.
For people unfamiliar with setting up Linux systems, it may be easier if you just go with a system like Ubuntu which has fairly recent pre-built packages (although not the latest - I don't think any Linux system keeps up with Mono releases) rather than wrestling with the learning curve of how to build everything yourself.
It is confirmed that in the near future Xamarin will support Linux and provide binaries (mono and mainline applications) for Debian and Centos derivatives, and their are already packages for Debian and Centos derivatives for technical preview. So cheers and no more pain of compiling and even parallel mono installaions.It can not get more easy than this. Check here

Best setup for Linux development from Windows? [closed]

Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 5 years ago.
Improve this question
What's the best setup for developing Linux apps from a Windows workstation? Right now I'm connected via SSH to our Linux development server and am using Eclipse, forwarded over SSH via PuTTY, to the public domain version of Xming running on my Windows workstation. It works, but it's not great; Eclipse's response times are far from snappy (noticeably worse than Eclipse running natively on my much slower Windows workstation), I can't resize some dialog boxes, and I haven't figured out a good way to reconfigure my fonts.
Is there a better setup available?
Edit: This is for C/C++ development.
Options for Linux on Windows:
Tools Only
Given you're using Eclipse I'm going to assume you want a full IDE, but if you can get by with just the GNU/Linux tools, there are a few choices.
cygwin gives you a bash shell with lots of tools, including an X11 server. This has been around awhile and is mature.
msys is a smaller, lightweight alternative to cygwin.
GNU utilities for Win32 is another lightweight alternative. These are native versions of the tools, as opposed to cygwin which requires a cygwin DLL to fake out its tools into thinking they are running on Linux.
Linux in a Windows Process
There are several packages that will run Linux as a Windows process, without simulating an entire PC as virtualization does. They use Cooperative Linux, a.k.a. coLinux, which is limited to 32-bit systems. These don't have the overhead of virtualizing, and they start up faster since you're not booting a virtual PC. This is a little more on the experimental side and may not be as stable as some of the virtualization options.
Portable Ubuntu
andLinux
Virtualization
Virtualization software lets you boot up another OS in a virtual PC, one that shares hardware with the host OS. This is pretty tried-and-true. There are nice options here for taking snapshots of your Virtual PC in a particular state, suspend/resume a virtual PC, etc. It's nice to be able to experiment with a virtual PC, add a few packages, then revert to a previous snapshot and "start clean".
VMWare
VirtualBox
VirtualPC
In my case...
Sounds like your environment has different performance characteristics, but here's my situation: I started out with Eclipse on my Windows laptop (doing Rails development), found this sluggish, and switched to using putty to ssh into a fast Linux box. I do my editing via an emacs running on the Linux server, displayed on Windows using Xming. Or I use native emacs on Windows, editing the files shared via NFS. The latter is slower in my environment due to sluggish saves.
When working from home, I ditch X because it is too slow with remote clients, and just run emacs -nw within a putty window. I then use GNU screen so that I have multiple "windows", and so that I can easily resume where I left off if my network connection flakes out.
The best approach that I've found is to:
keep your code portable
develop natively on your desktop
verify any OS dependencies (minimize these as much as possible)
deploy to your target regularly, test & debug there
I know that this isn't a direct answer, but using an IDE for development through X is painful with most of the free tools. The only way that I've been productive doing work this way was when I was running a UNIX-like on my desktop so X was native. If you are going to use this approach, try a commercial X solution on the desktop.
Other than that, consider ditching the IDE and doing your development and debugging via SSH, a terminal editor (e.g., vi, pico, ee, emacs), make/ant, and gdb.
The best approach for you is going to be driven by your programming language and the type of application you're developing. If you are doing GUI applications, then using X might be the only approach that is acceptable. If you are doing back-office/daemon development, then the SSH and terminal approach will probably work though you probably want to get really comfortable with either vi or emacs.
EDIT: just noticed that you are doing C/C++ development. Consider using a cross platform framework if you aren't already. Using something like Qt, APR, ACE, or Poco should make it possible to natively develop under Windows with a deploy/debug step to your Linux environment.
For development I usually use a Linux virtual machine on my Windows box. It will probably send Linux users running to the bathroom to wash their hands, but I do all of my development in Visual Studio, and I have a custom Visual Studio plugin that invokes G++ through the virtual machine and pipes the output into the VS output window. With a quick change of a Combo box I can build and test for Windows or Linux.
An easy to setup option would be to run Eclipse natively in windows but deploy the code via a Samba share on the Linux machine (which you can mount as another drive) (or SSH/SCP if SMB is not an option) and then run it there via SSH console.
Another easy to setup option is to simply develop on Linux via freenx or a similar tool instead of a full blown X session, check this answer: https://serverfault.com/questions/11367/remote-desktopping-from-windows-to-linux/11372#11372
The other options (Virtualization, Linux running inside windows, Cygwin) are indeed valid but have their drawbacks, like being more machine demanding, harder to setup, or not equivalent enough to the actual linux environment, but may very well be worth your while if you have the machine and the scenario justifies their use.
Doing everything on the Linux side will always have some drawbacks
if your machine is Windows.
I personally have a Linux box where everybody else has Windows and
do Windows dev inside a VM, but it has costed me a lot of RAM and some network setup pains.
I find coLinux tremendously helpful when developing on Windows for Linux, it's basically a linux system running in parallel to your Windows OS (i.e. as a service) and can be configured to simply show up on your LAN, basically like a virtual machine does. Also, it's much more full featured than CygWin, and its performance is really remarkable - I can easily run non-trivial stuff under coLinux, and still run simulators at 90+ fps.
Also, coLinux can be easily set up to run X11 and window managers like gnome/KDE, so that you can for example use something like vnc to access your linux desktop.
Cooperative Linux is the first working free and open source method for optimally running Linux on Microsoft Windows natively. More generally, Cooperative Linux (short-named coLinux) is a port of the Linux kernel that allows it to run cooperatively alongside another operating system on a single machine
. For instance, it allows one to freely run Linux on Windows 2000/XP, without using a commercial PC virtualization software
such as VMware, in a way which is much more optimal than using any general purpose PC virtualization software.
(source: colinux.org)
There are multiple solutions, I'd recommend No. 1
A VM (Virtual Machine) running a flavor of linux as a guest operating system inside Windows. Start with VirtualBox which is free.
To make managing it easier you can use a tool like Vagrant. Vagrant is a tool for building and managing virtual machine environments in a single workflow. With an easy-to-use workflow and focus on automation, Vagrant lowers development environment setup time, increases production parity. So you code in your Windows PC and compile/run the application on a Linux system using Vagrant. Vagrant is free! Similar tool: Docker can be used too. For this setup you can use any IDE, I'd recommend VSCode its quite handy for C/C++ with intellisense but Eclipse should work too.
Web based tool like Nitrous.io which is discontinued, but you can host your own open-source version of the Nitrous IDE called Nitrous Solo which lets you host your own instance of the Nitrous IDE on your preferred cloud provider.
Windows 10 provides provides Windows Subsystem for Linux, try using that to compile and run your project. This requires a 64-bit version of Windows 10 Anniversary Update or later (build 1607+).
Cygwin / MinGW are popular bash tools for Windows, they might be able to compile and/or run your application.
Cygwin might be helpful.
I've done what you want to do for exactly the same reason: full control over the output (you're having font issues with your current solution) and much slower Windows machine than the remote Linux development box.
Most answers are bogus: having a "Linux development environment" is not just "having an IDE". It's about having the whole Un*x power at your fingertips.
Is it a local or remote Linux server? bandwith issues? Because on a LAN, even an old 100 MBit/s LAN, FreeNX flies. How's the load on that Linux server?
Setup the free FreeNX on the Linux system, install the free FreeNX client on the Windows machine and bingo, you've got your Linux development environment at your fingertips.
FreeNX is much more efficient than VNC, it's night day (VNC is actually pretty bad perfs wise, even compare to Windows's Remote Desktop... But FreeNX flies).
Regarding speed, a long time ago, I set up my main Linux workstation (it was a Pentium 4 / 2GB of memory back in the days) on which I was developing full-time using IntelliJ IDEA (another IDE), to serve a full X session (complete with a window manager etc.) that another developer was displaying remotely to... run another IntelliJ instance (and access all the Un*x niceties). It was on a LAN 100 Mbit/s and it was as if the app was local for the other developer.
Anyway, on today's hardware I cannot imagine how this could not work: I now have here a Core 2 Duo / 4GB of ram as my main desktop and a Gigabit LAN.
Such a setup was working perfectly 4 years ago, it would work perfectly today.
Now if you tell me you have bandwith issues or that the Linux machine you've got your account on is under heavy load or that it's not on the LAN, then things may be different...
How the younger developers who want a powerful Un*x system do it at the company I'm consulting for nowadays (that only has Windows desktops)? Most of them bring their shiny MacBook Pro and use that to develop ;)
I'm using xming as well and suffer from the same problems with Eclipse. Apparently, neither switching to cygwin makes it fast enough. Eventually I switched to developing in vim via xming. It doesn't take as much time as I feared to get used to all the key combinations, and the performance is absolutely smooth. Actually, now sometimes I use vim even when working natively.
Either a Virtual Machine with a Linux-based dev environment, or a local copy of some toolchain-agnostic IDE (e.g. Notepad++, with testing done via MinGW or CygWin as far as you can), or just write in Notepad++ and keep uploading to your dev machine and testing there, which is what I do.
You might try other X servers on Windows such as xwin32 and hummingbird. Note that these are commercial implementations.
Another solution is to install a VM server on your Windows box and install Linux on the VM. Options include VMware (non-free) and Microsoft Virtual PC (free download). VMware is much nicer than VirtualPC (64-bit support, more incentive to support Linux client OSes, etc.).
EDIT: In the last 13 years since this post was originally made, Cygwin/X (and Xming) has gotten a lot better. It's worth trying again. I now use it for my everyday work again.
You could take a look at setting up a svn server on the linux box and then using something like TeamCity todo a build on commit. You could write your code locally and do a commit when you want it to be compiled.
I don't know if there's a more modern route, but the standard way in my time was to run X Windows in Microsoft Windows, that way you can run any number of applications on your Ubuntu machine and control them and display them in Microsoft Windows
Check Check out.
You could try using any of the linux distros for windows, even windows-store have ubuntu, SUSE etc for windows and this could help reduce your coding efforts. This linux distros contain linux shell, kernel etc so you won't be needing linux system everytime debugging or testing your code.
You could also use Visual Studio Code which is far better and fast compared to eclipse and is even supported in linux and mac.
Check this for ubuntu distro on windows store.
Linux distros can also be downloaded from other sources but microsoft urges to use the one from Windows-Store.
Use Linux! I usually have the other problem: developing win under linux.
There is no reason for not doing so: I have win running on a virtual box now almost all the time.
Linux comes with a lot of development tools.
The problem is:
is it a graphical interface?
If no you will have no problems as soon as your code STD/portable.
(X allows you simple stuff too but for an nice application today you need a bit more.)
If Yes then you will have a lot of problems when you actually port the code
on the running platform.
Is it supposed to be portable/exchangeable between linux and windows?
if not, just develop on the native OS. Way less pain. You have Eclipse for both
platforms. Even if you think to port the code on a later stage,
just do the work for one first.
I developed a couple of graphical application under linux which are actually right now
used only under windows. My recipe is: GTK/GNOME. I made it running with cygwin and mingw.
But I guess that Qt has the same usable environment too.
My code went on win with no changes!
[ok.. a couple of touching on file paths... but was a bug..]
There is no way to develop under win and hope to be running on linux unless you are sure
not to use any win libs. That is: in a graphical application almost no chance. Or a lot of
checking... Or you will not be using any win facility. Forget Visual Studio.
Check indeed wine and the winehq pages.
Unless the problem is another, like: using team sharing facilities, or svn or whatever.
Which is not a code development problem but a bit more on the organizational side.
Bottom line:
It is way easier to port a free code on win then a proprietary code on the free market.