Coderepo is an open source polyglot algorithm and data structure code repository. If you're impatient, check out coderepo.io.
Coderepo aims to have thousands of fully-worked, tested, and immediately runnable code modules in an exhaustive variety of languages.
We currently have code in the following languages:
- C
- C++
- Clojure
- Coffeescript
- Go
- Haskell
- Javascript
- Python
- Ruby
- Rust
Coderepo can be used in several ways, which are outlined here. We describe each usage in more detail later in the readme.
- Standard Usage
- The website, coderepo.io, provides a try-in-the-browser interface.
- Contributor Usage
- The "do_everything_from_scratch" command will create an Ubuntu 14.04 Vagrant box and install everything you need to start contributing.
Stack Overflow and Rosetta Code are both amazing, but they often provide code snippets without context. Those snippets require an implied version of an interpreter and various tough-to-satisfy prerequisites. Even if your setup is perfect, the code often contains small typos and logical errors that cause bugs because the code isn't automatically tested.
Having tests for the code has several advantages: First, the tests act as a client of the code, ensuring that it works. Secondly, the tests can demonstrate to the audience how the code is intended to work, acting as a specification. Finally, the tests serve as an example of how testing can be done in the language.
Unlike most learn-in-the-browser sites, we don't aim to provide an approximation of languages in a REPL in the browser, rather we aim to provide real languages running real programs with real tests in the browser. We accomplish this by running the real programs in Docker containers.
Coderepo aims to be very easy to contribute to.
There are only two prerequisites, Virtualbox and Vagrant.
Once you have those installed, you can run:
./do_everything_from_scratch
It will download an Ubuntu 14.04 base box, install all the software, then run the tests.
In runner/runner_containers, add a directory consisting of a Dockerfile,
installer, and runner. See any of the existing runner containers as examples.
Let's say you add the following runner container: elixer_1_0_0
To build it
cd runner; ./docker_build elixer_1_0_0
If you just want to add a shared code template that runs in one of the pre-existing runner containers, you can just copy one of the directories in the shared_code_templates directory and make changes to it.
For example, you could
cd shared_code_templates; cp -a kmp_in_python stack_in_python
Then you could make some changes in stack_in_python and run it with:
cd runner/build_and_run; ./run_one stack_in_python
If you want to use your newly created runner container, though, you'll need to understand the directory structure a bit more.
In shared_code_templates add a directory consisting of all the files you
would like to be able to run in the runner container, and also a Runner file
which specifies the name of the runner to use. In this case, let's make
a shared code template called binary_tree_in_elixer. The Runner file
consists of elixer_1_0_0
Once you've done that, you can run the code in the container with the following command:
cd runner/build_and_run; ./run_one binary_tree_in_elixer
Many test harnesses use different reporting formats. Coderepo gets around this problem by using custom regex matchers for each reporting format, allowing us to create a meta-harness that helps us ensure that all the tests for all the code examples pass. Having such a harness in place allows us to release new versions of the full stack quickly and without fear of regressions.