The one-paragraph verdict

The best parts of this book are the deep dives into Python’s dunder methods — __new__, descriptors, __getattr__, the numeric and container protocols — where Lott explains not just the syntax but why the data model behaves the way it does. That material is hard to find written this clearly, and it changed how I design classes. The trouble is the title oversells the focus: large stretches drift into SQLite, shelve, REST, and unit testing, which are fine topics but dilute the object-oriented thesis. The Packt editing also shows in places — uneven prose and a few examples that feel padded. Read it for the protocols, skim the rest.

Who should read it

Intermediate Python developers who already write classes but don’t fully understand the data model behind them — this is where it earns its keep. Library and framework authors designing hierarchies or custom protocols will get the most value. Beginners should not start here, and anyone wanting a tight, modern OO-design treatise will find the scope too diffuse.

Where it earned its place

The protocol chapters paid off directly when I built the document model for Archives. Treating chunks as objects with well-behaved __len__, __iter__, and equality semantics — rather than passing dicts around — made the chunking pipeline far easier to reason about and test. The book’s framing of “make your objects act like Python expects” was the concrete idea I kept reaching for.

Skip it if…

You want a focused, opinionated guide to OO design, or you’re past the intermediate stage and already fluent in the data model. At that point the protocol chapters are review and the rest is reference material you’d sooner look up.