Feature request: Allow customizable task hashing #6683
Description
New feature
During workflow development, minor changes to a process script (e.g., adding debug echoes, refactoring commands, or commenting lines) frequently invalidate the task cache, even when inputs and outputs remain functionally identical. This forces unnecessary re-execution of tasks on every -resume run, significantly slowing down iterative development.
It would be very useful to have a way to customize how task hashes are computed, particularly to optionally exclude certain components (like the process script content) from the hash calculation in specific scenarios.
Use case
- Rapid prototyping and debugging of pipelines.
- Prevents re-running long tasks just because a cat or echo was added for inspection.
- Complements existing caching mechanisms without breaking reproducibility in final runs.
Suggested implementation
Introduce a new process directive (or extend the existing cache directive) that allows users to control which elements contribute to the task hash. Examples:
process myProcess {
cache { task ->
// Return a custom object/map that contributes to the hash
[name: task.process, inputs: task.inputs] // explicitly exclude script, container, etc.
}
// ...
}