Records, ordering, and invariants

The idea

A dataclass generates routine record methods from annotated fields, helping a program expose meaningful names instead of positional conventions. Type annotations document intent but do not automatically validate incoming runtime data. A frozen dataclass prevents normal reassignment of its fields; it does not recursively freeze a list stored in a field. Prefer immutable field values when the record must remain dependable.

Sorting is a separate policy from record storage. Python sorting is stable, so equal keys retain their earlier relative order. A tuple key expresses primary and secondary rules directly, and sorted returns a new list. Explain ties explicitly rather than depending on incidental input order when a report needs a unique order. Validate boundary data before constructing records that internal code treats as trustworthy.

from dataclasses import dataclass

@dataclass(frozen=True)
class Entry:
    name: str
    score: int

rows = [Entry("Bo", 8), Entry("Ada", 8), Entry("Cy", 5)]
print([r.name for r in sorted(rows, key=lambda r: (-r.score, r.name))])

['Ada', 'Bo', 'Cy']

Negating score gives descending numeric order, while the name provides an ascending tie-breaker. sorted leaves the original list order unchanged.