Open mwaxmonsky opened 1 month ago
@climbfuji @nusbaume @dustinswales @peverwhee @mkavulich I noticed this in a PR review recently and then found it in another PR so this has been around for a bit.
The fix is pretty straight forward to add xmllint
to the environment but is this a needed part of the ccpp_capgen.py
execution process? All of the tests pass so currently it is clearly not needed but I am wondering if we should change this behavior or remove it if it isn't needed as a part of the capgen execution.
We can go in either direction but I usually lean more toward requiring it (at least in this case as it looks to be checked every time we use the capgen script) and failing if the dependency is not met and adding a flag to explicitly turn it off (ie, --disable-xmllint
) or vice-versa (making it an explicit opt in ie, --lint-xml
).
Any thoughts/opinions on if/how we should change the behavior?
I like the idea to make this an option --enable-xmllint
or --disable-xmllint
. Whether the default should be on or off, I am undecided. If I put on my operations hat, I'd say the default should be off. Folks preparing for operational implementation can have it turned on, but once a suite (that never changes) is in operations we don't need the check.
Description
Current capgen unit tests all pass but there is an error message hidden in the log output
Steps to Reproduce
Run any of the current capgen tests in the CI
Additional Context
N\A
Output
Sample log file excerpt from https://github.com/NCAR/ccpp-framework/actions/runs/11220321952/job/31188229511: