aboutsummaryrefslogtreecommitdiff
path: root/internal/openmeteo/testdata/pollen.json
diff options
context:
space:
mode:
authorLukasz Kasprzak <lukas@labunix.xyz>2026-08-25 15:54:53 +0200
committerLukasz Kasprzak <lukas@labunix.xyz>2026-08-25 15:54:53 +0200
commit391fc97fca932a50557d8e0f07ea2dfa8135900a (patch)
tree3df2978f60be0cc6b0a5f108beb39a4850e98b20 /internal/openmeteo/testdata/pollen.json
parent02503d2fd8b6c6c4e765f1f59661e7c337ebe971 (diff)
downloadprognosis-391fc97fca932a50557d8e0f07ea2dfa8135900a.tar.gz
prognosis-391fc97fca932a50557d8e0f07ea2dfa8135900a.zip
ci: fail the build when documentation falls behind
Adding a flag and forgetting the man page was the kind of thing only a reader would catch, and the README had already drifted: it documented seven of fifteen flags, claimed three cross-compilation targets where the Makefile builds six, and its example output predated the humidity column. Flag definitions move into defineFlags(), so the test enumerates the same set run() does rather than a hand-copied list that could drift in its own right. usage() gains a writer so its output can be captured. Tests then assert that every flag reaches -h, the README and the man page; that every key written into the generated config is documented; and that every column name is explained. The man page check normalises roff's \- hyphen escape first -- without that, every multi-word flag looks undocumented when it is not.
Diffstat (limited to 'internal/openmeteo/testdata/pollen.json')
0 files changed, 0 insertions, 0 deletions