Open
Description
The current repeatPolicy has implicit behavior that changes based on configuration:
condition+expected: Repeats until condition matches expectedconditiononly: Repeats while condition returns exit code 0exitCode: Repeats while exit code matches
This implicit behavior is confusing. For example, to monitor while a service is "RUNNING", you have to think backwards and set it to repeat until it's NOT running.
Solution
Add an explicit mode field to repeatPolicy:
# Clear intent: repeat WHILE status is RUNNING
steps:
- name: monitor-service
command: check_status.sh
output: STATUS
repeatPolicy:
repeat: while
condition: "${STATUS}"
expected: "RUNNING"
intervalSec: 30Examples
# Wait UNTIL service is ready
repeatPolicy:
repeat: until
condition: "nc -z localhost 8080"
intervalSec: 5
# Retry WHILE failing
repeatPolicy:
repeat: while
exitCode: [1]
intervalSec: 10
limit: 5
# Monitor WHILE healthy
repeatPolicy:
repeat: while
condition: "curl -f http://localhost/health"
intervalSec: 60Implementation
- Update
repeatfield type inRepeatPolicystruct toRepeatMode - Accept constants values:
whi 4ECD leoruntil - Maintain backward compatibility by inferring mode when not specified:
repeat: trueor unspecified (empty) -> defaults towhile
- Add integration test
- Update the json schema
- Update the documentation on repeat policy
Benefits
- Clear Intent: No more mental gymnastics to invert logic
- Better DX:
repeat: whileis self-documenting
I’d appreciate any thoughts or suggestions you might have about this feature.
Metadata
Metadata
Assignees
Labels
No labels