Inherited Asymmetric Descriptor Detection & MRO Type Specialization in Microsoft Pyright

Engineering objective

System challenge

When Python classes inherit properties with asymmetric getter and setter types across complex method resolution orders (MRO), Pyright assumed symmetric type bounds on property access, emitting invalid diagnostic errors on valid assignments and reads.

Constraints

Working within real limits

• ✕Preserve full compliance with Python typing PEPs (PEP 484, PEP 544, PEP 681)
• ✕Zero performance regression in Pyright type check latency across multi-million line codebases
• ✕Handle multiple inheritance and diamond MRO patterns deterministically
• ✕Pass full Pyright Jest, tsc, and regression test suites

Approach

How it was built

1. 01

   Specialized property getter types across class MRO hierarchies during static attribute evaluation.

2. 02

   Differentiated asymmetric getter/setter signatures during member access and mutation.

3. 03

   Authored regression coverage covering complex inheritance, descriptor protocols, and protocol conformance.

4. 04

   Resolved upstream merge collisions and validated with 100% clean test passes.

Technical validation

Implementation evidence

Observable engineering evidence. Discipline: MERGED UPSTREAM

Public evidence

Review the upstream contribution

Outcome

What was delivered

• Eliminated false-positive diagnostics for inherited asymmetric descriptor properties

• Received 4 maintainer approvals across core Pyright team, marked mergeable

• 100% pass across complete Pyright test battery

Technologies

Stack and tools

Previous and Next Engineering Systems