Ideas how to test+benchmark my custom datastructure?


(J Ohn Stuart) #1

At work we process large data and every data item (numbers) obeys some rules. They are not random numbers as they obey some rules like: all in each category share prefixes, sum of digits has special properties, etc.

I have written a custom map like data structure, that works on our data. It is better than the standard map, because it uses this additional info that we have about our data.

Here comes the hard part: convince my boss and my colleagues to use my data structure.

My plan is to write lots of unit tests for it (for all corner cases I can think off), then also benchmark my data structure against the standard one (map).

Do you have any ideas or suggestions of what tests/benchmark I could do in order to prove that my custom data structure is better for us ? I will only get one shot to make a presentation, so my proof should be as good as possible.

Please share your thoughts on what tests/benchmarks a custom data structure should have, in order to convince you to use this one instead of the already existing (in public packages) similar one.