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=v0.6.0 2026-08-18T11:24:48Z 2026-08-18T11:24:48Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T11:24:48Z urn:sha1:f979aedcc8a11d2c5a175064526a6821e11c2c0c 2026-08-18T11:23:36Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T11:23:36Z urn:sha1:5d3e201c21427ec54ff3d58bade04e40cb621ae9 2026-08-18T11:23:23Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T11:23:23Z urn:sha1:d727d07472172e3308ebd1706476a6bbf7ec1306 Both listed check and new-overlay among the commands but neither told a reader how they fit together, which is the part that makes them useful. The overlay sections now carry the four-step loop -- new-overlay, edit, check, run -- and state what check does not do, since its name invites a stronger reading than it earns. Also documents what the format now permits: citations and layer optional, the signed Easter_offset, the negative nth counting from the end of the month, and the legal values of each of the six required fields, which previously appeared only in the shipped example. 2026-08-18T11:20:50Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T11:20:50Z urn:sha1:e48fa4a2d315ac2b3832611ff48ed2e986e7b49e Two mandatory fields made optional, parse errors that stop naming kernel source files, and a feedback loop -- colitur check and colitur new-overlay -- for a file the test layers deliberately cannot vouch for. 2026-08-18T11:16:53Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T11:16:53Z urn:sha1:49494c3aedf3f1b40c6ee36cb4de5615f097274c Writing a local calendar had no feedback loop. An overlay is applied, not validated -- that stays true, and the five test layers still cannot vouch for a user's file -- but before this the only way to learn whether yours did what you meant was to generate a year of output and grep for your own slug, and the only way to see that a directive matched nothing was to notice a warning scroll past among 365 lines. `check` loads each overlay, applies it to the real shipped calendar, and reports the directive counts, the slug each one targets, and any directive that found no target. It exits 2 when a file fails to load or a directive matched nothing, so it composes into a Makefile or a pre-commit hook rather than merely being readable. It is applied to the SHIPPED calendar and not to an empty layer on purpose: against an empty one every Suppress would fail trivially and the check would be worthless. It answers three narrow questions -- does the file parse, does every directive find its target, what does the merged result contain. It does not validate a calendar against the rubrics and cannot, and both the help text and the man page say so rather than letting the name imply more than it does. `new-overlay` prints a starter to stdout for redirection, rather than writing a file where it likes. Every value in it is a placeholder that will appear in `day` output if left unedited, so a half-finished overlay is visible rather than silently inert, and it documents the three date shapes and the legal values for each closed field inline. load_ef_layer now returns its diagnostics instead of printing them: day and readings still want them on stderr beside a year of output, while check wants them on stdout, attributed to the overlay that produced them, and counted. Printing at the source made the second impossible. The cram test round-trips new-overlay through check rather than pinning the template line by line -- editing its prose should not fail a test, but a syntax error in it still must. 2026-08-18T11:13:40Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T11:13:40Z urn:sha1:8d0ab547895460956e36b66b47a299b02a3474f0 A user-supplied overlay is the only sexp this engine reads that a human writes by hand, and two of Celebration.t's eight fields carry nothing such an author can meaningfully supply: citations is always empty for a local feast, since citations come from the rite's lectionary and never from calendar data, and layer merely repeats the overlay file's own id. Requiring both made the commonest first mistake -- omitting them -- fail with 'lib/kernel/celebration.ml.t_of_sexp: the following record elements were undefined: citations layer', which names a source file the author will never open and does not say what to write instead. A minimal local feast needed 12 lines, two of them noise. Overlay.load now fills each field only where ABSENT, so an explicitly stated value always wins: an overlay may legitimately name a layer different from its own id, and defaulting must not silently overwrite that. A test pins both directions. Deliberately scoped to overlays. Layer.load, which reads the shipped sanctoral, is untouched and stays strict -- that data is the project's own, every field of it is asserted by tests, and a missing one there is a defect rather than a convenience. Parse failures also stop naming kernel source paths: the five prefixes that actually reach a user are rewritten into the vocabulary of the file being edited, and anything unrecognised passes through verbatim rather than being reworded into something possibly wrong. 2026-08-18T10:57:18Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T10:57:18Z urn:sha1:084ff22997af208715656924d9fc23bf2ec10050 colitur has been downstream of lectio since its data was bootstrapped from it. These are the fields where that relationship should now invert: eight changes, each annotated with the colitur allow-list entry carrying its rubric. Deliberately a PATCH and not a regeneration. colitur carries 205 of the 327 Polish names lectio ships, so regenerating lectio's ini from colitur would silently drop 122 of them; colitur also uses a different slug vocabulary for Passion/Holy week and different Paschaltide week numbering. What colitur is authoritative for is the adjudicated fields, and only those are emitted. Comparing the two datasets field by field also caught a counting error in colitur's own register that no assertion in this project was watching for: six sanctoral entries carry subject = Lord, not the four recorded, because the regex that counted them truncated on the two longest Polish names. The RG 112(b) reachability measurement was re-run against all six and is unchanged at 0/0; both docs are corrected. 2026-08-18T10:43:36Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T10:43:36Z urn:sha1:4d143d762713783d35b2b1d20f8a21d79884dafe They have been carried as a bare 'unbuilt' for weeks. The reason is not neglect, and it differs per clause -- now measured across 1583-9999 rather than asserted. (c) cannot fire in this architecture at all: Temporal_ef.temporal returns one office per day, so two de Tempore candidates never coexist. Nothing to build. (b) is already produced by RG 16(a), which is implemented: 0 days in either direction, because a Lord feast takes an occurring Sunday's place with no commemoration, and no Class1 Sunday shares a date with any of the four Lord-subject entries. (d)'s saint half has no candidate pair. Three apparent pairs are distinct saints sharing a forename -- recorded so the false positive is not rediscovered -- and the one real pair, agnes and agnes-secundo, is 7 days apart and co-occurs 0 times. (b) and (d) are data-unreachable rather than architecture-unreachable, so an overlay can make either live. That is the entry_14_movable_band shape. Left as an explicit open decision rather than silently resolved. 2026-08-18T10:41:26Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T10:41:26Z urn:sha1:7ee8aaf865203aa2134d0af661cb011f87650fd2 CLAUDE.md still described it as architecturally blocked, and that claim was being carried forward into task lists a release later. It was true when written and was superseded by the very next task: the blocker was that no (month, day) pair could anchor an Easter-relative trigger, and Date_spec gained Easter_offset in v0.2.0. The entity has shipped since then as an ordinary Add in adjustments.sexp -- 1 981 days across the domain carry it. Also recorded the general lesson, since this cost a stale item twice: a blocker phrased as an architectural impossibility was really a claim about one type's expressiveness, and it dissolved when that type grew a constructor. 2026-08-18T10:37:36Z Lukasz Kasprzak lukas@labunix.xyz 2026-08-18T10:37:36Z urn:sha1:d5386c2121cfd5c6b73b573aa7db05b8ef21dff0