Iterators

Why this matters

itertools, pandas' .iterrows(), and every streaming parser (csv.reader, json.JSONDecoder.raw_decode) rely on the iterator protocol. Staff engineers prefer generators over materialized lists for log processing and ETL pipelines because they compose lazily and hold O(1) memory over billion-row streams.

Common pitfalls

• Iterating the same iterator twice — it's exhausted after one pass; wrap in list() or use itertools.tee.
• Forgetting to raise StopIteration in a manual __next__ — infinite loops and hung for blocks follow.
• Using .iterrows() on large DataFrames — 100× slower than vectorized ops; reach for .itertuples() or vectorization.