jeffspies / benchmark Goto Github PK

Currently benchmark follows unit test and pytest in treating methods beginning "test" are the ones that are "special". I think that benchmark should change from "test" to "benchmark" as the special start of method name.

Currently benchmark follows the unittest philosophy of a script being a valid Python script with everything self-contained. pytest takes the view of getting rid of the "boilerplate" by having a special command required to run the script, i.e. the script is not a Python script but a pytest script.

Is it feasible for benchmark to support both approaches or should, like unittest and pytest, the two approaches be two separate benchmarking frameworks?

i was wondering if anyone has given thought of expanding this to cover resource usage like memory, disk and so on. i think this would be useful for that purpose: https://docs.python.org/2/library/resource.html

forking may be necessary to properly monitor each test invidually in this case. a separate data structure should also be built parallel to results to track other resources.

the benchmark/statistics.py file has the following notice right now:

Copyright (c) Maurice H.T. Ling <[email protected]>
@see: Ling, MHT. 2009. Compendium of Distributions, I: Beta, Binomial, Chi-
Square, F, Gamma, Geometric, Poisson, Student's t, and Uniform. The Python 
Papers Source Codes 1:4

what is the actual license of that file? is it free software?