Measure the performance impact of Debugger, the debugging API of Mozilla's SpiderMonkey JavaScript engine, when running the Esprima parser on its own source.

You will need:

• A SpiderMonkey shell executable, built with debugging disabled and optimization enabled.
• A copy of the Esprima source code. You can simply point it esprima.js from a git checkout: git@github.com:ariya/esprima.git

Edit those into bench.sh.

Things to try if you want to bring the standard deviation down:

• Make sure CPU throttling is disabled. You don't want the later benchmarks to run more slowly just because the processor got hot.
• Look out for noisy processes like web browsers, mail readers, artificially intelligent text editors, ...
• Disable networking, perhaps?
• Run directly from the console, to avoid GUI futzing altogether?