Is your feature request related to a problem? Please describe.
We currently are checking the performance of our methods in a very informal case-by-case way. This mean that we do not have accurate measurements of performance gains/loses when we change things around, nor can we easily detect if a change impacts performance negatively.
Describe the solution you'd like
A common way to measure and compare performance in the Python world is to use asv tests. These are used all around the Scientific Python ecosystem, for example in NumPy, SciPy, scikit-learn, or Pandas.
We should try to integrate this type of tests in order to work in performance topics more effectively. As opposed to unitary tests, these tests are not usually run in CI, but only under demand.
Is your feature request related to a problem? Please describe. We currently are checking the performance of our methods in a very informal case-by-case way. This mean that we do not have accurate measurements of performance gains/loses when we change things around, nor can we easily detect if a change impacts performance negatively.
Describe the solution you'd like A common way to measure and compare performance in the Python world is to use asv tests. These are used all around the Scientific Python ecosystem, for example in NumPy, SciPy, scikit-learn, or Pandas.
We should try to integrate this type of tests in order to work in performance topics more effectively. As opposed to unitary tests, these tests are not usually run in CI, but only under demand.