deterministic OCaml engine to compute and validate liturgical calendars for multiple rites, template-driven output to year 9999 https://git.labunix.xyz/colitur.git/atom?h=main 2026-08-26T21:53:58Z 2026-08-26T21:53:58Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-26T21:53:49Z urn:sha1:90584d87e763789808329847c46dbaa8e22dca47 Parse.verse_range's [last] endpoint gains an optional chapter (Parse.verse_end: { chapter : int option; verse : verse_num }), so a hyphen range whose two endpoints lie in different chapters ("1 John 1:5-2:2") can be represented at all. Rejected: a bare [int option] living alongside [last] as a second field on verse_range -- that would let 'a chapter with no verse' exist as a constructible value. parse_part now splits a part's leading "chapter:" at the FIRST colon only (not every colon), so the verses side can itself carry a second colon from a crossing range. parse_range detects a crossing by checking whether the range's right-hand side contains ':', and skips the same-chapter descending-range guard for that case (a later chapter is always "ahead", whatever its own verse numbers are). Handles the compound shape too -- a crossing range followed by further, same-chapter verse references in the same comma list ("Matthew 9:35-10:1,5a,6-8") -- since those trailing pieces parse as ordinary bare verses/ranges, unaffected by the preceding crossing. Render's one_range renders a crossing [last] through the same chapter_verse template one_part already uses for the part's own leading "chapter:verses", so a style that reconfigures the chapter/verse separator renders a crossing endpoint in that same convention rather than a hardcoded ':'. This closes the W4 known-wrong: 41 (now 49, after an intervening Second-reading extraction) of the OF lectionary's citations printed unconverted, every one this exact shape. test_citation_coverage_of.ml's pinned residual is now empty and asserted exactly, over the full 1725-field data/of/lectionary.sexp population, including the round-trip check (parse -> render -> parse structural equality). test_citation.ml gains direct parse-suite cases for the basic crossing, the compound shape, a mid-list crossing, a crossing with a sub-verse letter, and a malformed-crossing rejection. data/of/lectionary.sexp is regenerated via its own generator (tools/bootstrap_lectionary_of.ml, whose own embedded header text is updated to match); only comment lines change, confirmed by diff -- no lectionary entry differs. test_lectionary_of.ml's whole-file SHA-256 pin is updated to match. EF is unaffected: data/ef/ is untouched since v1.0.0, and a direct byte comparison of `colitur day`/`colitur readings` for 2026, 1583 and 9999 against a git-worktree build of 1c0137d is identical on all six outputs. 2026-08-26T11:30:02Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-26T11:30:02Z urn:sha1:9cc763c169a6ce4ebbdbed506c5eb2dad1f9f3c1 35% of OF citation fields (259/730 on colitur readings --rite of 2026) printed unconverted -- 1 John renders on 2 January but not 3 January. Two independent causes, both in the citation/siglum path, neither in the OF data itself: 1. Book.table only ever surveyed the three EF citation-bearing files, so 345 references citing a book no EF file happens to use (Job, Ruth, Judges, 1/2 Samuel, 1/2 Chronicles, 1/2 Maccabees, Baruch, Ecclesiastes, Habakkuk, Haggai, Nahum, Zechariah, Zephaniah, Deuteronomy, Amos, Micah, Lamentations, Ezra, Joshua, 2/3 John, Jude, Philemon, plus "Isaiah"/"Jeremiah"/"Ezekiel"/"Malachi"/"Mat"/ "The Acts"/"Tobit"/"Song of Solomon" spelling variants of books EF already knows) failed as "unknown book". Added as a new, separate of_lectionary_table rather than folded into the EF-surveyed table: none of these 25 new books is attested in the EF's own scans, and lang/la.ini's own header refuses to fabricate an uncited Latin title, so they are resolvable (parse + render, falling back to their own English spelling) but deliberately excluded from Book.all -- test_lang_coverage.ml's la.ini-completeness promise is preserved exactly for the ids it already covered, not silently weakened. 2. Parse's grammar could not read a verse number carrying a lectionary sub-verse letter ("11a", "1bcde") at all -- the dominant remaining failure shape once (1) was fixed. verse_range now carries a verse_num { n; suffix } on each boundary, PRESERVED through rendering rather than dropped (dropping would silently lose real precision the source text carries). Chapter numbers are untouched (nothing in the data ever attaches a letter to one). A third, subtler bug surfaced by (1): registering "jude"/"philemon"/ "2 John"/"3 John" exposed Parse's existing "leading comma-number is a chapter" heuristic misreading a single-chapter book's bare verse list ("Jude 17,20b-25") as chapter 17 -- a wrong PARSE, worse than the previous safe "unknown book" failure. Book.is_single_chapter now tells Parse to skip that heuristic for the four one-chapter books and default to chapter 1. Residual, honestly enumerated rather than forced to zero: 41 distinct references (of 1540) are hyphenated ranges crossing a chapter boundary ("2:29-3:6") -- a Parse.t shape verse_range/part do not represent, a type restructuring deliberately not attempted this task. Pinned exactly by the new test_citation_coverage_of.ml, both directions (a new failure or one of these 41 starting to convert both go red), and disclosed in data/of/lectionary.sexp's own regenerated provenance header (tools/ bootstrap_lectionary_of.ml now runs the same parser at generation time and names the count and the set). Verified EF-unaffected: git diff v1.0.0..HEAD -- lib/kernel/ lib/rites/rite_ef/ data/ef/ is empty, and `colitur day`/`readings` output for 2027 is byte-identical against the pre-fix binary. 2026-08-20T17:03:02Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-20T17:03:02Z urn:sha1:1da70dc7ac03fe33fb92b172a0e26932764170d6 The branch was RED and reported green. `dune test` exited 1: test/cli.t pinned the pre-fix output `kings_1 19:3-8`, which the previous commit had already fixed to `1 Reg 19:3-8`. The gate command piped dune through `tail`, so it reported tail's exit status, and cram prints its diff BEFORE the alcotest summary, so the two lines shown were the passing ones. Verify with `dune test; echo $?`, never through a pipe. A style file's own `book` key was unreachable. sigla_book resolved against a hardcoded "abbr" and the result was applied unconditionally, so the documented `[sigla] book = full` could never win. Render gains book_string, and the style's own value is now the default that a flag or config overrides. The unit test pinned style_of_fields correctly while the wiring defeated it. `lang --check` filtered the reference set to the celebration prefix, so a file with no [bible] section at all reported a clean bill of health -- contradicting both the reason the keys change was made and lang.ml's own comment. It now reports missing book names too. The token test missed a FOURTH citation-bearing file: adjustments.sexp writes citations as `Set_citation`, not `(reference ...)`. Its 16 citations all parse, so nothing was broken, but nothing was checking. The first attempt at this fix read the file and extracted NOTHING -- the marker stopped before the opening quote, so every payload was the part label -- which is recorded in the code rather than left as a trap. Also: colitur-config(5) claimed a trailing period the data does not carry, and two la.ini scan quotes silently corrected OCR damage ("Ionae 3, I - I O", "Epistolse") while presenting themselves as verbatim. Both are now marked as corrections. 2026-08-20T13:09:49Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-20T13:09:49Z urn:sha1:a2cbb6b85e79fbc59b0879362c0f853757d51c07 A style is a set of format strings, so punctuation convention is data. Values are unquoted here rather than in Overlay_ini: that parser trims every value and is shared with overlays and [defaults], so teaching it about quotes would change behaviour this feature has no business changing.