Serialization More than pickling
summary
A concrete comparison of serializers using an ebird API example where a BirdCounter object holds 266 birds and construction takes milliseconds but re-running costly queries can take hours The talk actually benchmarks json pickle dill messagepack and marshmallow shows json ~12 KB pickle faster smaller dill slower comparable size explains dill works for CTF as it reconstructs objects across kernels demonstrates pickle executes arbitrary code by running system commands and recommends schema tools like protobuf or marshmallow when needing versioning while noting messagepack small and fast for attribute-only serialization
AI-written summary · about
chapters (4)
related talks
30mProcessing large JSON files without running out of memory
30mPython Serialization and the Hallows of Parallel Computing
51mMarshalling Pickles
33mData Class Serialization The Right Way
28m"Performant Python"
40mBeyond Pandas: The Great Python Dataframe Showdown
37mProcessing large JSON files without running out of memory
44mWorking Around the GIL with asyncio