I work on a windows machine but I ssh into an AWS linux box to do most of my work. I'm trying to figure out a way I can use my local tools (specifically Spyder) to open up files (python files) from the server so I can edit them in the IDE and then run them on the server. (Note: its not just for code, I also want to be able to open csv's I generate, etc...).
Many modern IDE's provide this option through SSH, looking at the documentation for Spyder, it seems to refer to this configuration as a 'remote kernel'.
I do not use Spyder, but I use a similar configuration to do development with Visual Studio Code on my laptop with remote execution over SSH to an EC2 instance. In VS Code it is referred to as Remote SSH, you open a folder on the remote server as your VS Code workspace.
Remote development on Linux from Windows is easily doable via SSH.
However, what about the other way? I need to build and debug my Visual C++ application on Windows, but I want to work on a Linux system.
Cross-compiling via MinGW doesn't work because of MSVC-specific libraries
Ubuntu on Windows is a good start, but I'd like to work on a real Linux system
RDP/VNC or something like that doesn't help either, because than I'd work on Windows again
So does a virtual machine
Maybe something like Powershell on Linux + SSH to the Windows Powershell?
I regularly develop Visual C# applications remotely from Linux, not MSVC for the most part, but, like you, I wanted to find a way to build and debug Windows-targeted applications and libraries on remote Windows machines without working directly in the box using RDP, Visual Studio, etc.
It's difficult to answer this question without more information about the development and debugging tools you prefer to use on Linux for the types of applications you develop. I'll try to provide a general overview and update the answer for details you add about your workflow.
Cygwin, similar to MinGW's MSYS, provides a Unix-like environment for Windows. Most importantly, Cygwin, unlike MinGW/MSYS, includes an implementation of the OpenSSH server that enables us to connect to the Windows box over SSH from Linux (or any other device with an SSH client, really). We can install the sshd package using Cygwin's setup utility. After connecting, Cygwin drops us into a Bash shell by default. With this capability, we can:
Execute remote commands and scripts over SSH.
Edit files using our favorite *nix command-line text editor (Vim, Emacs, etc.)
Mount remote filesystems locally using SSHFS (if Windows shares are unavailable).
Forward or tunnel ports if needed.
The availability of a general-purpose shell makes almost anything possible. We can execute batch files, PowerShell scripts, and native Windows executables from Cygwin's shell environment in addition to Linux scripts and Cygwin programs.
For example, we could run msbuild from the SSH session command line to build our VC++ application or we could configure our local GUI editor or IDE running in Linux to execute msbuild over SSH when we click the "build" button.
We could set up a similar environment in recent versions of Windows using the Windows Subsystem for Linux ("WSL", Bash on Windows). I personally prefer Cygwin for greater portability and ease of configuration. Cygwin's sshd can run as a Windows service, and, as an established project, Cygwin integrates very well with Windows systems (user accounts, filesystems, Windows APIs, etc.).
Working with Code
We can choose from several workflows depending on our tools and comfort-level with the command-line:
Completely text-based—all work performed through the SSH session
Use local tools on files mounted in a remote filesystem
Use local tools and synchronize files
I use the first approach. I'm a heavy Vim user, so I connect to Windows machines over SSH to do my work on the command-line using the tools and environment provided by Cygwin. The availability of tools typically found on Linux simplifies many tasks that are hard to do from the default Windows console. We can write shell scripts to automate tasks that Visual Studio might normally do for us. For example, I wrote a wrapper script around mstest that reads the XML test results and outputs them in a format that's easy to read in a terminal.
If we prefer to use a GUI editor or IDE, we can mount the remote code locally so tools can read and write files as if they were part of the Linux machine's local filesystem. We likely still need to use SSH to execute commands needed to build the projects, but many editors allow us to configure this command as the project's "build" action.
Sometimes a remote filesystem is too slow for effective editing. In these cases, we can synchronize files between the Linux development machine and the Windows host using a tool like rsync or the editor's "upload on save" feature (over SFTP, for example), if available.
Debugging
Everything works pretty well until we try to find a way to debug our applications. As of now, there is no reasonable substitute for Visual Studio's debugger when working with Visual C++ projects. We can debug managed C# applications running on the CLR using MDbg, but no comparable tool exists for C++ programs.
We can try to use gdb (from MinGW, Cygwin, etc.) for basic, low-level debugging of native binaries, like reading memory addresses, but the debugger does not yet support reading Microsoft's debugging symbols, so the debugging experience is very limited. Microsoft began documenting the PDB format a couple years ago, so we may see some compatibility in the future. Even so, it will take a long time to produce a satisfactory alternative to Visual Studio's excellent debugging tools.
For debugging, RDP is currently our best—and probably, only—option. For a more native-feeling experience, we can run Visual Studio using rdesktop (or other RDP client) and seamlessrdp to create a single-window RDP session of the Visual Studio IDE instead of a full desktop which integrates with whatever window manager we're using on Linux.
Sometimes we can get around launching a full Visual Studio debugging session for simple debugging scenarios by adding tracing to our application that outputs values to the console or to a log file. In many cases, this is faster than starting the debugger anyway.
We can also try to use Eclipse's CDT debugger configured for the Visual C++ toolchain. This may enable us to perform remote debugging using an Eclipse instance on the Linux machine. I have never tried this approach, and I expect there may be some issues when the application is linked against Microsoft's libraries.
I don't know all your requirements, but maybe you could use a gdbserver on Windows (from MinGW) and remote debug from VSCode on Linux - or any other environment you like. You can find more details in this post here. (Watch out, VSCode prevents you from running gdb unless it’s signed as mentioned in the first link.)
There is also a Native Debug VSCode extension that could be helpful.
Another solution I can think of is to use Visual Studio Online (free for small teams up to 5 persons) as build server.
As you have said, the other way around is pretty easy and nowadays even officially supported by Visual Studio 2017.
Most probably, the VS remote debugging tools for Windows wont be helpful for you.
So I have a remote Centos server running a small app and I have a git repo set up on that machine. To date, I've been running git from that centos server via ssh terminal, including doing some merging via mergetool, specifically vimdiff. This is...okay. Not great. A terminal editor split into four panes is vaguely terrifying.
What I would like to do is handle the merges via a Windows machine and some friendly GUI based merge tool, but I don't want to clone the repo to the windows machine. I want the windows machine to perform the merge on the Centos machines files directly.
I suspect this is possible, but I'm not having much luck with google. A link to instructions would be fine.
Any help?
I have a set of machines, a mixture of Linux and Windows Boxes.
I hav set up rsync to pull from the Linux Machines to a Linux Server box.
I am trying to accomplish the same using cwRsync, to pull to the Linux box from the windows machines. I have downloaded the free version from https://www.itefix.no/i2/content/cwrsync-free-edition and also I have downloaded CopSSH. I have managed to install CopSSH fine and I am able to SSH between the Linux and Windows hosts no problem using keys rather than passwords.
However, for the life of me I can't get this cwRsync working, I've googled the matter to death, and your meant to unzip the directory, configure the environment settings in the batch file then install it. However, there is nothing to install it with! and the reason it isn't working is because it needs to install a windows service for it to run.
Any help would be much appreciated!
As described at itefix web page for the free edition, it allows to initiate rsync from your Windows machine, i.e. client functionality only (push data). Server functionality allowing you to set up an rsync server on Windows to pull data from it is not a part of the free edition.
I have a working copy of a repository on a Windows web server. If I update/commit the working copy using TortoiseSVN (1.7.6) or the command line (version 1.7.8) "svn update" on my Windows PC I have no problems.
However if I am on my Linux Subversion server (where the repository lives), accessing the working copy through a CIFS mount, without fail within 10 updates/commit the file .svn/wc.db will have become corrupted. Sometimes it is with the message
svn: E200030: sqlite: database disk image is malformed" and sometimes it is a message like "svn: E155010: Pristine text 'd9a9a3ee5e6b4b0d35bfef95601890afd80709' not present
I can clean up the corruption every time it occurs, but clearly I don't want to keep having to do that - I am struggling to work out the cause of the problem.
My Linux Subversion server is also using version 1.7.8. What could be causing the problem?
It is probably because the internal representation of the files or in the database are slightly [or much] different on a Linux and a Windows machine. Make your own copy on the Linux machine, and commit from that into the central repo, then pull it down on your windows machine.
In general, it's never a good idea to access the same copy of a version controlled repository from two different machines.
I use SVN for web work, and that's how I work - I do nearly all my work on a Linux machine, but I do have a repo on my windows laptop.