Keynote Speaker – James Powell
summary
Argues that replacing a monolithic scipy.optimize.newton API with an explicit generator yields concrete gains: fewer combinatorial keyword arguments, far fewer tests (five modalities force 32 cases), and easier custom stopping logic implemented with itertools takewhile and islice or a collections.deque for the final value. Demonstrates using numpy and scipy for evaluation, cites xray vs pandas indexing as a specific API ambiguity, and shows the generator approach both simplifies code and preserves flexibility.
AI-written summary · about
chapters (6)
- 0:00Start
- 13:27Design.
- 14:55convenience layer
- 16:15Judgement?
- 16:43example
- 23:00unambiguous delineation of the "steps".
related talks
43mWhy do I need to know Python- I'm a pandas user
01h 10mKeynote 2: Robust Python and Why It Matters
49mGenerators Will Free Your Mind
01h 00mKeynote: Python now and in the future
29mI wish I learnt that earlier!
52mKeynote: Learning programming and science with Scientific Python
01h 18mKeynote
01h 15mEducational framework for Black Box optimization methods design