Benchmarking of OO hierarchies

The following benchmarks have been constituted and gathered by several researchers, e.g. Karel Driesen and Jan Vitek.
Yoav Zibin has been my main direct source.
I have also added some new benchamrks, e.g. SmartEiffel and PRM (thanks to Jean Privat and Floréal Morandat).

Format

The format is a variant of the cou format, a LISPish format readable by Common Lisp.

Each hierarchy is a file constituted of a single list of class descriptions---here is the variant, since usual cou format contains separate class descriptions, not a single list.
Each class description has the following form :

(<class-name> <list-of-super-classes> <list-of-attributes> <list-of-methods>)

All lists have the usual LISP syntax.

Of course, attributes and methods are presumed to be non-static.

Here is the directory of all benchmark files.

Extended formats

In recent papers (ICOOOLPS 2010, JOT), we used a more complete code abstraction, with an extension of the cou format.

At the moment, there is only a single benchmark in this format, the PRMc benchmark.

Simulation platform

My simulation platform is written in Common Lisp and CLOS.
It computes various parameters (numbers and sizes) according to many implementation techniques and produces LaTeX tables used in various publications.
I would be happy if people want to use and extend it. Unfortunately it is somewhat such a gaz plant that I don't know if it is usable by somebody else than me.
Anyway, if you are interested, contact me.