There are a number of corner-cases in core ansible semantics that don't
appear to be nailed down. I don't know about anyone else, but we certainly
seem to have a knack of depending on those without realising it.
I also note there seem to be a fairly high level of bug reports about
regressions introduced with each new ansible release.
Is there interest in accepting regression tests that document expected
ansible semantics? I imagine that they'd basically all be a simple
directory containing a top-level playbook together with expected output.
If there's somewhere to put these I'll be happy to start supplyig them
Cheers,
jan