A framework for benchmarking shell-commands. Written in Haskell.

The idea is to provide a simple, but precise, easy-to-use environment where it is possible to launch benchmarks and collect execution time.

stack tool is required. After you have install it in your system:

git clone https://github.com/fusiled/Skater
cd Skater
stack build

The library heavily leverages Turtle for spawning shell commands. The execution time is collected with the criterion package.

stack exec Skater-exec <path-to-skater-context> <path-to-benchmark-file>

A basic test example can be launched with:

stack exec Skater-exec examples_json/default_context.json examples_json/benchmark_example.json

You can write you own JSON files by looking at the simple examples provided in the examples_json folder. If you want to add different type of Benchmark you can extend the data Benchmark inside src/Skater.hs

• Improve JSON parsing. The JSON parsing is very simple. The idea is to use generics for making the system more flexible (Removing a Map from the source code in Skater.hs)
• Implement usage of stderr stream (a Turtle limitation?). It seems that with Turtle it is possible to use the stdout stream. It could be possible to try to hard-code in the command string the redirection to /dev/stderr, but it doesn't look very nice.
• Implement test-cases

This is my first work written in Haskell. Please be patient if everything is not performed in the Haskell way. For any feedback, please contact me.