Should nccmp in functional testing compute the maximum difference across all data values for regular ANTS?
- Dominant language
- Python
- Stars
- 7
- Forks
- 10
- Avg merge
- 1d 19h
- Merged PRs (30d)
- 5
Description
# ✨ Feature Request
## Requirements
We truncate the number of different values output in our rose ana testing to just [5 values](https://github.com/MetOffice/ANTS/blob/06fdecaeb3850d06cb3f2e50c6d12cae2040843a/rose-stem/app/rose_ana/bin/ants_compare_files.sh#L18) for brevity. Unfortunately, this also means that statistics - including maximum difference - are only computed for the first 5 values. We should consider using all different values for the statistics, even if this means including all different values in the nccmp output.
## Context
If we opt to do this for core ANTS, we should also have similar fixes in contrib, and UG-ANTS/UG-Contrib.
Contributor guide
Research direction
Start with rose-stem/app/rose_ana/bin/ants_compare_files.sh, especially the five-value output limit linked in the issue. Run the functional comparison and inspect how nccmp statistics are derived from the truncated values. Done means maximum-difference statistics use all differing values, with corresponding treatment considered for core ANTS, contrib, UG-ANTS, and UG-Contrib.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- shell
- Domain
- testing-qa
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100