Add targeted coverage tests to maintain >80% code coverage minimum:
- ExtractInput error paths (14 tests): Type mismatch errors for Result,
Option, Vec, HashMap, Arc, and primitive types. Directly exercises
error formatting code that was previously uncovered.
- DagRunner::get() type mismatch (6 tests): Covers the type mismatch
error path when attempting to retrieve task output with wrong type.
Tests various type combinations to ensure robust error handling.
- TaskBuilder conversions (9 tests): Covers the From<TaskBuilder> to
TaskHandle conversion for unit-type tasks, ensuring ID preservation
and correct handling of mixed task types.
Also fixes timing test threshold from 10ms to 50ms to account for
tarpaulin instrumentation overhead while still validating parallel
execution.
Coverage increased from ~78% to 82.04% (above 80% minimum requirement).
Enable tasks to accept ANY type implementing Clone + Send + Sync as
dependencies, without requiring users to implement ExtractInput trait.
The #[task] macro now generates task-specific extraction logic inline
via a new extract_and_run() method. This leverages the macro's perfect
compile-time knowledge of parameter types to handle extraction for:
- Zero parameters (unit type, source tasks)
- Single parameter (any Clone + Send + Sync type)
- Multiple parameters (tuples, awaited concurrently)
Previously, only hardcoded primitive types and specific stdlib types
worked. Custom user types required manual ExtractInput implementations.
Now custom types work automatically with zero API changes.
Changes:
- Add extract_and_run() method to Task trait (internal/hidden)
- Generate inline extraction in #[task] macro based on parameter count
- Remove ExtractInput bound from DagRunner::add_task()
- Retain ExtractInput for internal task_fn() test helper
- Add comprehensive custom type tests
No breaking changes - purely internal implementation improvement.
TechnoPorg
swaits