PikaPython v1.14.0 Release Announcement

We are pleased to announce PikaPython v1.14.0. This release focuses on Python 3 syntax compatibility, runtime correctness, VM performance, fatal-error visibility, and threading support.

New Features and Improvements

• Python 3 Compatibility: Improved short-circuit expressions, chained comparisons, unpacking, comprehensions, default arguments, variable arguments, and keyword argument binding within the existing syntax subset.
• Exception Semantics: Typed except now matches exception types and parent classes correctly. Nested handler exceptions propagate outward, and except ... as alias preserves the actual exception type.
• Threading Support: Added the threading package and improved Lock/RLock behavior and tests.
• VM Performance: Reduced redundant function, argument, keyword, and object-path lookups. The Fibonacci workload uses approximately 10.1% fewer dynamic instructions across two Callgrind-verified optimization rounds.
• Fatal Error Diagnostics: Stack exhaustion and other fatal resource failures now report capacity, requested size, and execution state before entering the platform panic path.
• Release Tooling: Added consistent validation for runtime versions, package indexes, CMSIS Pack metadata, Git tags, and release commits.

Bug Fixes

• Fixed crashes and silent failures caused by malformed imports, assignments, comprehensions, and unsupported expression forms.
• Fixed incorrect behavior in grouped comparisons and short-circuit expressions.
• Fixed default parameters whose value was False.
• Fixed positional, keyword, starred, and variadic argument binding errors.
• Fixed typed exception dispatch and parent-class exception matching.
• Fixed nested exception handlers losing the outer try state.
• Fixed iteration and slicing errors that could leave invalid runtime state.
• Fixed excessive argument handling so invalid capacity requests are rejected before writing outside the VM stack.
• Improved error propagation and heap cleanup after failed runtime calls.
• Fixed Linux CI isolation and benchmark publishing failures caused by generated files and optional dependencies.

Performance Improvements

• Optimized recursive global function lookup.
• Removed redundant unqualified host-object path scans.
• Reduced repeated keyword constant-pool scans.
• Reduced argument-binding and unpack-validation overhead.
• Streamlined successful VM error-stack checks.
• Reduced core Flash usage in several optimization rounds without adding persistent RAM.

The Fibonacci benchmark showed an accumulated 10.1% reduction in dynamic instructions. This is not presented as an exact wall-clock speedup because the current benchmark export does not yet retain complete repeated-sample statistics.

Module Releases

Validation

• Release metadata, runtime headers, CMSIS Pack metadata, package indexes, and the annotated tag are consistent.
• Linux Release core compilation passed.
• Python 3 exception, recursive lookup, and error propagation tests passed.
• Development-cycle Linux regression and Valgrind checks completed without heap leaks.
• A fresh clone of v1.14.0 passed release consistency checks and core compilation.

Known Issues

• optimize_speed exception tests may still be sensitive to GTest registration and startup allocation layout.
• Some Linux REPL tests still depend on unstable fixed log indexes.
• GitHub Pages benchmark exports do not yet include complete repeated-sample variance statistics.

Special thanks to all contributors and users who reported compatibility, runtime, and embedded-system issues.