GCC and Clang on Windows with Docker

If you're looking to get some work done with gcc or clang but you're on Windows, fear not - there's a way to do it without regrets.

I played around with GCC and clang for my exception handling post. Typically I work on a Windows machine, and prior experience has taught me that getting GCC and Clang working just right can be a bit of a mess.

So, I installed the trusty Docker for Windows and got going. If you, like me, need a refresher every time you're about to do something interesting, follow along!

Installing Docker Desktop for Windows

Instructions on how to install Docker are available here. I often reach out for Chocolatey, where you can get set up with a simple choco install docker-desktop.

Making sure things work

If you've just installed docker or if you haven't used it in a while, you may want to make sure you're properly set up.

• Open up a command prompt.
• Run docker --version to check your version.
• Run docker run hello-world to make sure it works.
• Run docker run --interactive --tty ubuntu bash to get an interactive Ubuntu container. Run hostname or whatever, then exit

So far, so good. If you didn't get this far, you'll need to get this sorted out before you can continue.

Now let's actually build something.

Setting up the file

Open up notepad and paste the program you want to build, for example as ex.cpp. You can seve it to %TEMP% if you want.

#include <stdio.h>
void main() {
  std::string message("hello, world!");
  printf("%s", message.c_str());
}


Docker containers aren't great at user interface, so you'll want to edit the files from something richer. Even notepad looks good when all you have is a CLI.

GCC

• Open up a console, change into to the directory with the file - say %TEMP%.
• Run docker run --rm -v %CD%:/usr/src/myapp --interactive --tty gcc bash, and run the following.

cd /usr/src/myapp
g++ -o ex ex.cpp
./ex


The switches, explained:

• --rm: automatically remove the container when it exits
• --interactive: keep STDIN open even if not attached
• --tty: Allocate a pseudo-TTY
• gcc: this is the image name to run
• bash: this is the program to run

To avoid the cd, you can also specifify the working directory in the docker run command with -w /usr/src/myapp.

The gcc image comes from here. You can look at the various tags to get specific versions, and you can look at the linked Dockerfile for any one to see how this works. The scripts are not trivial, but they're not too hard to follow along either and show the details of how things are set up.

Clang

Clang doesn't have a ready-made image in docker, but we can simply spin up an Ubuntu image and install it.

• Open up a console, change into to the directory with the file - say %TEMP%.
• Run docker run --rm -v %CD%:/usr/src/myapp --interactive --tty ubuntu bash

apt-get update
apt-get install clang
cd /usr/src/myapp
clang++ -o ex ex.cpp
./ex


Clang, better

Now, re-installing Clang every time is a bit of a pain. Instead, we can create an image to our liking, and then start from there.

Create the following plain text Dockerfile in an empty directory:

FROM ubuntu as BASE
RUN set -ex; \
    apt-get update; \
    apt-get install -y clang;


Now run the following in that directory from a command prompt: docker build -t myclang:v1 . (note the trailing period).

Go back to your temporary directory, and build with this:

docker run --rm -v %CD%:/usr/src/myapp -i -w /usr/src/myapp --tty myclang:v1 bash

You can verify you have things set up properly with clang --version.

This RUN command is there to make it easy to add other things to your images; best practices for simple apt-get and a few other RUN commands is available here.

Happy compiling!

Tags:  coding cpp tutorial

Home