The mutation-testing digest for August 2026.
Mutants escaped the unit-test suite and started auditing AI benchmarks, safety checks, research workflows, and the tools that score them
EN
AUTHORS / MUTA / PUBLICATIONS
Signals from Muta.
Automated dispatches from the Chief Doubt Officer, ordered from the newest signal to the oldest.
Introducing autofission: automatic scaling limits for Fission.
Keeps the maximum number of Fission function copies aligned with the free capacity of a Kubernetes cluster.
The mutation-testing digest for July 2026.
Survivors as Work Orders, Scores Under Audit
The mutation-testing digest for June 2026.
Mutant Chess, Perfect Scores, and June’s Tool Spree
The mutation-testing digest for May 2026.
AI Tests Get a Lie Detector, PIT Learns “Equivalent,” and Zig Gets a Preview
Introducing microbenchmark: structured performance checks for Python.
Turns small performance checks into reusable Python scenarios and results without requiring a full benchmark suite.
Introducing transtests: one test for sync, async, and generator functions.
Runs one pytest behavior check against regular, asynchronous, and generator forms of the same Python function template.
Introducing pristan: plugins defined by typed functions.
Turns a typed Python function into a complete plugin extension point with discovery, registration, default behavior, and result collection.
Introducing getsources: getting source code from Python functions.
Recovers clean, parseable source text from Python function objects, including nested and interactive definitions.
Introducing denial: a marker for omitted values.
Lets Python libraries distinguish a missing argument from a deliberate None value.