Programming C++ on Linux vs Programming C++ on Linux VM in Windows

Question

Is there any difference between learning to program C++ on Linux as opposed to learning via using a Linux VM, on Windows?

My initial guess is that there is no difference - the VM simply acts as if it were a normal Linux OS?

I am doing this to become particular with C++ programming on Linux, including the Linux kernel, how Linux works, IPC, sockets, shared memory, pipes etc.

Answer 1

Probably the biggest issue you will run into with a VM will come when you are writing multi-threaded programs, or doing IPC on shared data. Unfortunately because of the way most VM's work, you can end up masking timing issues that would cause crashes on a "real" machine natively running the Linux kernel, or at least un-desirable behavior. For instance, a data-race that may appear benign on a VM because it doesn't crash during run-time could cause, because of the inherently faster timing on a physical machine, a true data-race that may be very hard to debug. Another example of timing-related issues that could occur might be where you have multiple processes writing to a pipe with a single reader. Because of the speed of a VM, you might observe behavior where every writing process is able to atomically write their entire payload into a pipe, even if that payload is larger than the guaranteed atomicity of PIPE_MAX ... if you programmed for that type of behavior, on a real-machine you could end up with a big surprise as anything over PIPE_MAX gets interleaved with other processes writing to the pipe.

So in the end, the fact that you are able to observe a lack of process data corruption or crashes from potential data-races while doing multi-threaded programming or shared data IPC on a VM does not assure that your program is actually data-race free, or that it will not crash on a physical machine where the timing of interleaved operations will be much faster. The speed of the VM could simply be masking those issue for you.