The checker isolated XDG_CONFIG_HOME and not XDG_DATA_HOME, so the two
Ruthenian examples were verified against the MCI menaion this machine has
fetched and CI does not. They passed here and failed there, which is the
checker working: a reader running them after a fresh install gets the common
menologion, and that is what the README now shows.
Both examples still make their point. The commemorations are worded
differently; the fast, the obligations and the contrast between them are
identical either way.
The screenshot was a picture of the Python CLI: it predated the formats, the
tradition in the header and --typikon, and a picture cannot be diffed against
the program anyway. Every example is now a console block of real output.
tests/readme-examples.py runs each one and diffs it, which found four things
wrong in the process of writing them: the hero example printed today rather
than a fixed date, so it was wrong for any reader on any other day; the
Ruthenian example depended on whatever the reader's config remembered; a
tradition block held two commands and was checking neither; and the JSON had
a hand-collapsed array that the program does not emit.
It also found a real bug. --typikon could be turned on by the config file and
there was no way to turn it off again from the command line, so
on this machine showed the Typikon's discipline with no way to ask for the
law. --no-typikon now does that.
The Rust example is a test rather than prose, for the same reason.
funkpower