The CSV and JSON reporters should write to files, as the html reporter does

[jdmarshall]

I cannot think of a good use case for trying to parse CSV or json from the console where it is mixed with output from the suite.run() operation. These should go to files so they can be programatically processed.

[RafaelGSS]

By default, if you report to CSV or JSON, you can simply do:

$ node --allow-natives-syntax mybench.js > output.csv # or
$ node --allow-natives-syntax mybench.js > output.json # or

[RafaelGSS]

But adding an option to write to file is ok too. But, it should be optional.

[jdmarshall]

That won't suffice for CI workflows. Every unit testing framework - except Mocha - writes both to stdout and to an output file unless otherwise instructed not to. Plus any output from the benchmark itself would break your output file format.

I'm stuck on another problem at the moment and wondering if instead the generation of the report might need to be split to a separate method that returns text instead of either a file or stdout traffic. So I might then just write the file out myself instead of modifying bench-node.

But I will say that mocha is widely considered frustrating for making hoops you have to jump through to get ./output and stdout to see the same suite information.