For example, the full name of the standard python interpreter is `CPython`, however we would name it `python`, after the main binary which it provides.
In the example of NodeJS, we would call this `node`, after the main binary.
This script should download sources, compile sources and output binaries. They should be dumped into the current working directory, removing any files which aren't required in the process.
5. Create a file named `compile`, containing bash script to compile sources into binaries. This is only required if the language requires a compling stage.
The first argument is always the main file, followed the names of the other files as additional arguements. If the language does not require a compile stage, don't create a compile file.
6. Create a file named `environment`, containing `export` statements which edit the environment variables accordingly. The `$PWD` variable should be used, and is set inside the package directory when running on the target system.
7. Create a test script starting with test, with the file extension of the language. This script should simply output the phrase `OK`. For example, for mono we would create `test.cs` with the content:
8. Create a `metadata.json` file which contains metadata about the language and interpreter. This simply contains the language name, as in the folder name, the version as in the folder name, aliases that can be used to call this package, limit overrides (if any) that can be used to override the default constraints and finally a dependencies map.
If it all goes to plan, you should have a file named `[language]-[version].pkg.tar.gz`, in this case you're good to go, albeit it is preferable to test the package locally as follows
```shell
./piston build-pkg [package] [version]
./piston ppman install [package]=[version]
./piston run [package] -l [version] packages/[package]/[version]/test.*