What You Got Is What You Got
summary
Argues that numpy pandas and networkx are powerful but hard to embed in production: ndarray are contiguous unboxed machine types so they give speed and simple semantics, while pandas DataFrame subclassing often decays to base DataFrame so inheritance breaks; composition forces endless API plumbing. Demonstrates a mesh optical network example with transceivers spans and amplifiers to show practical pain. Shows hashing-by-id is sometimes acceptable, proposes class-decorator construction as unexplored option, and claims there is no perfect proxy in Python because C extensions and VM edges leak through.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 2:15Python objects in production
- 4:41Restricted computational domains
- 6:25Data types in business
- 9:11Composition vs inheritance
- 14:26Extending APIs challenges
- 16:53Inheritance hierarchies fail
- 18:38Subclassing built-ins limits
- 21:48Better integration needed
- 24:56Python object model split
- 29:04Proxying in Python limits
- 31:51Future directions for Python types
- 32:33Q&A
related talks
01h 25mAdvanced Metaphors in Coding with Python
11mStaying Relevant in an Ever-changing Tech Landscape
30mEmpowered by Python - A success story
45mRecovering from Enterprise
20mState of Modern Programing
22mWhat Developers and Economists Can Learn from Each Other!
19mAI missteps as stepping stones
01h 55mFrom Novice to Data Ninja