# Exceptions for scripts/check_language.py. One per line:
#
#     <path suffix>::<literal fragment of the offending line>
#
# BOTH halves must match, and the key is the line's CONTENT, never its number: a number moves with
# any edit above it and would rot for a reason that is not a change in the finding. An entry frees
# the whole line. Blank lines and lines starting with # are ignored.
#
# An exception is for a line where the non-English text IS the subject, never for German that
# nobody got round to translating. Each entry says which.

# The casefold fixture: this pair is the DATA of the test. It is the difference between Python's
# casefold() and Java's compareToIgnoreCase, so replacing the eszett would delete the measurement.
src/epics_pv_mcp/services/olog_client.py::casefold() does
tests/test_olog_update.py::compareToIgnoreCase folds per
tests/test_olog_update.py::.txt"

# Non-ASCII audit probes: the encoding is the whole point of the probe, so the characters are the
# test input, not prose.
tests/test_safety.py::probe = "SIM:PS-01
tests/test_olog_write.py::probe = "OLOG_WRITE logbook=probe

# known-limits paragraph 9 documents the guard's own Title-Case false-positive class, so its
# example proper nouns (von Neumann, Weil, Des Moines) ARE the subject of those lines. The guard
# blocking its own documentation here is correct behaviour, and these entries are the sanctioned
# relief the paragraph describes.
docs/known-limits.md::spelled like a German function word
docs/known-limits.md::each trip the word signal, because the ALL-CAPS discard
docs/known-limits.md::The mechanical repair, discarding a lowercase
