| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The BVM Saturday office is now genuinely correct, so both comparators'
allow-lists move -- every count derived from the harnesses' own failure
output, never transcribed or guessed.
lectio differential (data/ef/expected-divergences.sexp, C17, new,
colitur): lectio builds no equivalent office at all, so it keeps
showing the plain season colour on every otherwise-unoccupied Class4
Saturday where colitur now shows white. Season, slug and rank all still
agree (the office deliberately reuses the ordinary ferial slug), so
Colour_f alone is the complete diff set -- gated on colitur's own
rank/weekday/colour rather than a date list, the same shape M2 used, at
416 rows over 2005-2050.
missalemeum oracle (data/ef/expected-divergences-missalemeum.sexp): M2
is CLOSED, REMOVED, the same shape as this file's own M4/M6/M7/M9/M12/
M14/M17 precedents -- not re-adjudicated to a different verdict, gone
because the colour divergence it named no longer occurs on any of its
former 22 dates. What is left on all 22 is Observed_identity_unresolved
alone (colitur's BVM-Saturday candidate is temporal-origin and
deliberately carries no English name) -- exactly M18's own shape, not a
distinct citation any more. Keeping M2 as a separate id would have kept
silently absorbing these rows under a stale "colour differs" citation
whose own reason had stopped firing (subset's own diff-set test admits
a SMALLER set than the one named) -- the same "count proving cardinality
where identity was required" trap this project's review process
watches for. M18's own expected_rows moves 373 -> 395 accordingly.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
RG 110(c), word for word: "quoties vero oratio unius Apostoli addenda
est ad modum commemorationis, huic orationi additur altera immediate,
ante omnes alias commemorationes." "Huic orationi" (dative, "to this
oration") refers to the trigger's own oration; "additur altera" means
the companion is added to it, i.e. follows it. "Ante omnes alias
commemorationes" governs the pair's position relative to other,
unrelated commemorations, not the pair's own internal order. The
previous build prepended the companion uniformly in both RG 110
shapes, citing this same clause for both -- correct for (a)/(b) (the
day's own oration, not a list member, has nothing to order against),
backwards for (c), where the trigger is itself a list member and the
companion belongs after it. Fixed in rg110_additions: shape (a)/(b)
still prepends; shape (c) now splices the companion in immediately
after its own trigger. Re-pinned the unit test that had asserted the
wrong order, and corrected both rows' comments to cite the clause that
actually governs each shape.
No layer in this codebase compares commemoration order: the lectio
differential never compares commemorations, the oracle sorts both
sides into a multiset, and the golden tests' own describe function
sorts its commemorations field too. Recorded as a new, permanent blind
spot in CLAUDE.md and the register, not merely fixed around -- the
admit_cases unit table is the only place order is asserted at all.
Re-measured the full blast radius against the same pre-change binary:
same 14,627 differing days, same four shapes and counts, confirming
the fix changed only the content of the 593 shape-(c) rows and nothing
else. Mutation-tested: reverting the splice back to a uniform prepend
reddens exactly the shape-(c) unit test; a new golden pin for a real
shape-(c) day (2004-02-22) stays green under the same mutation, living
proof that golden pins cannot see this axis, not merely a claim.
Separately, M19's own predicate checked a date and a diff shape but
not which candidate -- proved insufficient by fabricating a second,
bogus companion on 30 June and confirming the whole suite stayed
green. Fixed by requiring the admitted commemoration to be exactly
commemoration-of-st-peter, mirroring C16's own identity guard. Added a
golden pin for 30 June, the only assertion of that entry's identity
outside the allow-list.
Also: corrected a false claim in the task report that the electronic
transcription carries no corroboration of the Holy Name fallback --
its own table of contents does, at a line not checked on the first
pass. Added two further citations, both re-verified against the scan:
RG 17's own closing paragraph, settling the Sunday shape's Sunday-
exclusion at the rubric level rather than only in the propers; RG
124(c), which assigns red to 30 June by paragraph number rather than
by the generic apostle/martyr default. Noted the fragility of deriving
the Sunday shape's RG 109(a) privilege from a slug-substring match
rather than from a citation. Cosmetic: a stray blank line and an
indentation inconsistency.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
RG 110 (Caput XIV): "In Officio et Missa S. Petri semper fit
commemoratio S. Pauli, et vicissim... pro unica habeantur" -- in the
Office and Mass of either Peter or Paul, a commemoration of the other
is always made, uncapped by RG 111's own admission count. Both
photographic scans, full text including sub-clause (c) (a separate
trigger this register previously had not transcribed: the same rule
also fires when one Apostle's oration is admitted merely as an
ordinary/privileged commemoration, not only when it is the day's own
office).
Three real pairs in the 1962 calendar, all confirmed on both scans:
25 January (conversion-of-st-paul + peter), 22 February
(chair-of-st-peter + paul), 30 June
(in-commemoratione-sancti-pauli-apostoli + a new
commemoration-of-st-peter). The first two already existed as ordinary
sanctoral candidates but were wired into RG 111's ordinary, capped
admission contest instead of this rule's own uncapped one -- the
defect this closes. 30 June had no companion at all: a genuine gap
in lectio's own source data (confirmed directly against
tridentine-calendar.ini) and in missalemeum's own oracle output for
both fixture years, not only a colitur bootstrap miss, so colitur is
now uniquely correct on this date among the three engines compared.
Added via data/ef/adjustments.sexp's own Add directive, the
architecture's existing extension point for exactly this shape of
gap, since data/ef/sanctoral.sexp is generated and lectio must not be
touched.
Precedence_ef.admit gained rg110_additions/rg110_companion_slug,
layered on after admit's own four ordinarily-capped branches decide
their normal result: checks both the observed candidate and every
already-admitted candidate against the three-pair lookup and prepends
any companion found, uncapped and additional, never competing for
RG 111's own slot.
Measured full blast radius (1583-9999, pre-change binary from git
archive at 59fbda8 vs the fixed binary, 3,074,246 days each side,
14,627 differing days, all four predicted shapes, zero unclassified):
3,619 Holy Name fallback days (previous commit), 3,533 days Paul was
wrongly excluded from Chair of St Peter's own day by a competing
privileged feria, 593 days Chair of St Peter itself was admitted only
as an ordinary commemoration with Paul entirely absent, and 6,882 days
the new 30 June companion is added. The 3,533+593 figure supersedes an
earlier, narrower 852-day estimate that measured only the delta
between two older code commits, not the rule's full scope.
The lectio differential needed no change: it does not compare
commemorations at all, and this fix never touches season, slug, rank,
or colour. The missalemeum oracle allow-list needed three changes:
M12 removed (22 February 2027 no longer diverges in commemoration
count), M15's date list gained 2027-02-22 (now reaching identity
comparison for the first time, where it lands on that entry's own
pre-existing, unrelated limit -- one of the two commemorations is
temporal-origin and unnamed), and a new M19 for the 30 June gap.
Mutation-tested, both halves separately: disabling the admit-level
addition reddens the two new unit tests, the oracle's own M12-shaped
checks, and an existing golden pin (updated to expect the Paul
companion); removing only the 30 June data entry reddens solely the
oracle's new M19 count, confirming the code and data halves are
independently load-bearing.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
spot precisely
The full breakdown, precisely: of 730 days, 331 are resolved (330
matching, 1 mismatched -- M13, Joseph vs the Seven Sorrows), 399
unresolved -- 373 in M18's own bucket, 26 absorbed inside M1/M2/M3/
M16's own widened subsets where the axis fires alongside their
pre-existing citation. M18's own note previously stated only its own
373 without this fuller context.
Also names the blind spot the axis exists to close in exact terms,
not just "the observed day's identity": a TEMPORAL-origin observed
day silently replaced by a DIFFERENT temporal-origin observed day of
the SAME rank and colour -- exactly Holy Family's own shape, and the
reason every prior layer stayed green through the whole gap. States
explicitly what the count pin does and does not prove: it guards
against the 373/399 population moving silently in either direction,
but cannot say which specific day moved or why.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
layer_m_reason's own five entries whose diff shape was checked by
EXACT list equality (M1, M2, M3, M13, M16) lost their match once the
observed-identity axis widened every row's diff set two commits ago.
Each is widened the same way commemoration-identity was already
widened for M13 once before: the axis now ALSO fires alongside their
existing citation, same root cause, same id -- M1/M3 (temporal-origin
observed day, no name), M2 (same, plus five Paschaltide dates that
used to match cleanly and now surface for the first time), M13
(Joseph, sanctoral, now a genuine title MISMATCH, not merely
unresolved), M16 (the temporal Passiontide feria observed there).
M18 covers the remainder: 373 of 730 days whose observed celebration
is temporal-origin and so carries no English name at all -- the
overwhelming majority of the window, including Holy Family's own two
dates in it (2026-01-11, 2027-01-10), counted here rather than shown
as a false Matched. Gated on the diff SHAPE alone (this file's own
convention for a population this size, the same reasoning M2's own
title-substring predicate already uses instead of a date list),
verdict unresolvable -- the same fourth verdict M15 already
established for the identical limit on the commemoration axis.
Building an English name onto the temporal cycle is Plan 4's own
scope, register §6.
Teeth, per the brief: temporarily corrupted hilary's own English
name in data/ef/sanctoral.sexp (14 January 2026, an OBSERVED sanctoral
day, rank and colour left untouched) and reran. Reddened immediately:
2026-01-14: observed-identity-mismatch differ -- colitur=(rank=3
colour=w ...) oracle=(rank=3 colours=[w] title="St. Hilary" ...)
rank and colour agree on both sides -- exactly the shape the
pre-strengthening comparator would have slept through completely.
Reverted (git diff on the data file is empty) before this commit.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Layer 4 (test_oracle.ml, vs missalemeum) compared the observed
celebration's rank and colour and stopped there -- it never checked
whether the observed day is actually the RIGHT one. Holy Family (11
January 2026: oracle title "The Holy Family: Jesus, Mary & Joseph")
is rank 2/white on both sides purely by coincidence -- an ordinary,
unnamed Sunday and Holy Family share both -- so this layer has been
silently blind to the missing feast since it was built.
Adds Observed_identity_mismatch/Observed_identity_unresolved,
following the shape and honesty of yesterday's commemoration-identity
comparison (Comm_identity_mismatch/Comm_identity_unresolved): colitur
identifies its own observed day by slug, missalemeum by an English
title, and the bridge is Celebration.t.names, read directly off
LD.observed. That name is only ever set for a SANCTORAL-origin
celebration -- a TEMPORAL-origin one (an ordinary Sunday, a feria, a
movable named feast) carries none, so most of the 730 days in the
fixture window land in Observed_identity_unresolved, not a clean
match. That is counted and reported, never silently skipped -- the
same discipline this file already applies to an unresolvable
commemoration.
WIP, expected red: 399 days now show observed-identity-unresolved
with nothing in the allow-list naming it, and five of the six
existing M-series entries whose own diff shape is checked by exact
list equality (M1, M2, M3, M13, M16) lose their match now that this
axis widens their diff set. Landed on its own, ahead of any
production fix, so it is the regression net the following commits are
checked against rather than a change bundled in alongside the fix
they enable. The next commits widen those five entries' own diff
shapes and add the new bucket's own cited allow-list entry, then build
Holy Family itself.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Each was recorded as a genuine data gap "traced upstream into lectio,
unfixable without touching lectio". That constraint is gone: all five
commemorations (St Agnes secundo 28 Jan, St Eusebius Confessor 14 Aug, St
Evaristus 26 Oct, St Theodore 9 Nov, St Boniface Martyr 14 May) are now
present in the re-bootstrapped sanctoral.sexp, and each divergence stops
occurring in the 2026-2027 oracle fixture -- verified by re-running the
comparator and confirming zero rows for all five ids, not assumed from the
data change alone.
Removed rather than kept as dead 0-row citations, matching the M17
precedent already in this file: the count pin
(test_layer_m_counts_match_citations) now enforces that none of these five
divergences can silently reappear. Corresponding dead code (m4_dates
.. m14_dates and their layer_m_reason branches) removed from
test_oracle.ml; a short pointer comment left at each id's old location.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix round 1 on the commemoration-identity/tie-break work: the "genuine
same-band tie" M17 adjudicated as open was not a tie at all. RG 91's
table enumerates only "dies liturgici" (real feasts) -- entry 24's own
text is "Festa III classis, in calendario Ecclesiae universae
inscripta" -- so a Celebration.t.status = Commemoration_only candidate
(a saint reduced to a bare commemoration) has no row in the table to
begin with. Precedence_ef.band used to read rank alone, so it silently
lent such a candidate the same table entry as a genuine Feast of its
own rank; the calendarium's own 22 September row confirms the
distinction in its own notation ("S. Thomae de Villanova ... III
classis. / Commemoratio Ss. Mauritii et Soc. Mm." -- Thomas carries a
class number, Maurice carries none).
Fixed at the source, not by adding a status tier to compare_precedence:
band now checks status first, ahead of every rank-keyed branch, and
returns unclassified for any Commemoration_only candidate. M17 deleted
(not re-adjudicated) -- 22 September now resolves identically on both
sides, with no divergence left to name.
Blast radius measured independently (a temporary git worktree at the
pre-fix commit): 4451 days change across the whole 1583-9999 domain,
exactly 4 verified shapes -- 3576 x maurice-and-companions-martyrs ->
thomas-of-villanova and 282 x eleutherius -> philip-neri (admitted set
changes), 502 x appollonia/cyril-of-alexandria and 91 x dorothy/titus
(order only, independently confirmed against the calendarium's own
listing convention). Zero days change who is observed. Further swept
the whole domain for a genuine "two different candidates on the
identical real table entry" residual -- the shape RG 113 truly gives
no instruction for -- and found it empty: of 109201 same-rank ties
domain-wide, 83950 now resolve by band alone and the remaining 25251
are exclusively two Commemoration_only candidates tied at unclassified,
never two real feasts. The residual admit still breaks alphabetically
is real, but narrower than previously documented.
The "66 days" figure is reconciled, not merely disputed: 599 is the
tie population, 65 (whole-comparator: 67) is the real admitted-set
decision count within it (independently reproduced against the
pre-Task-B binary), 149 is order-only. Every one of the 65 real
decisions traces to the same Commemoration_only/Feast root cause as
Maurice/Thomas.
Also: validate.ml's admission fixed-point check recovers a
commemoration's real origin (by matching its slug against the day's
own temporal office) instead of reconstructing every candidate as
Sanctoral, since band -- unlike EF's own admit -- does read origin;
precedence_ef.ml/.mli's motivating example and blast-radius claims
corrected to no longer rest on the fixed Maurice/Thomas case;
test_identity_rank_corroboration's own population count corrected now
that 22 September (and 21 other Feast-status matches) became reachable;
its rank-agreement pin documents exactly what it does and does not
prove. Register (§4, §6.1, §6 checklist) and CLAUDE.md corrected in
place, including the RG 113 first-sentence gap this exercise surfaced
but did not close, recorded as a new open item.
275 -> 276 tests (with the exhaustive sweep), all green.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
test_oracle.ml (validation layer 4) used to compare commemoration
presence and List.length only -- cardinality, never which candidate is
admitted. A reviewer proved this vacuous by reversing admit's dignity
sort (the engine admitting the worst commemoration, an outright RG 111
violation) and finding every assertion in this file green across all
730 days.
tools/extract_missalemeum_oracle.py now also emits each commemoration's
own missalemeum id (field 9, commemoration_ids), parallel to the
existing title field; fixture regenerated from the same underlying
lectio snapshot (commit d7da4b0, re-verified, first 8 fields
byte-identical), read from a scratch copy so ~/git/projects/lectio was
never touched. SHA-256 pin and provenance updated.
The mapping: colitur identifies a commemoration by slug, missalemeum by
an English title string. No hand-built slug<->title table -- colitur's
own Celebration.t.names already carries the same English text for every
SANCTORAL-origin commemoration (bootstrapped from lectio, verified to
match missalemeum's titles character-for-character). This resolves
identity for every sanctoral-origin commemoration but cannot resolve a
TEMPORAL-origin one (Rite_ef.Temporal_ef never sets a celebration name)
-- that case is a new, distinct outcome (Comm_identity_unresolved),
counted and allow-listed, never silently skipped: 19 of 227
non-empty-commemoration days over 2026-2027.
Two genuine identity mismatches found and adjudicated (new allow-list
entries M15-M17, one is the unresolved bucket): M16, 27 March 2026, the
already-known unimplemented Seven Sorrows of Passion Friday office
(verdict missalemeum); M17, 22 September 2027, the RG 113 same-band
tie-break residual (verdict open, per the sibling admit fix). A new
corroborating check cross-references missalemeum's own id-embedded rank
against colitur's, split by status: Feast-status matches must agree
exactly (none reachable in this 2-year window, checked regardless);
Commemoration_only-status matches either genuinely agree or fit a known,
pre-existing, already-registered convention gap (colitur's bootstrap
rank vs missalemeum's own sentinel) -- not a title-matching defect.
Perturbation test performed and reverted: reversing admit's dignity-
based ordering (compare_precedence's real committed form uses band, not
dignity -- this reproduces the historical defect shape directly) turns
the oracle layer red: "no differences outside the cited allow-list"
fails with an unexplained commemoration-identity-mismatch on 2026-09-23
(colitur admits "St. Thecla" where the rubric and missalemeum both
require "St. Linus"). Reverted; diff against the prior commit confirms
no residue.
CLAUDE.md's "know what each layer cannot see" section updated: the
commemoration-identity gap it documented as open is now closed, with
the mapping's limits and the RG 113 tie-break finding summarised.
271 -> 273 tests (274 with the exhaustive sweep), all green.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
data/ef/expected-divergences-missalemeum.sexp's M6 entry and its matching
comment in test/test_oracle.ml both cited 1962-06-23,_SS_Ioannes_XXIII,
_Missale_Romanum,_LT.pdf's 14 August row ("Vigilia, II classis." --
nothing else) as if it were the calendarium itself. It is an electronic
transcription that silently drops commemoration lines the photographic
scans carry. missale-romanum-1962.pdf's 14 August row reads "Vigilia, II
classis, Commemoratio S. Eusebii Conf.", with its own proper following --
independently confirmed in the second scan, Missale Romanum 1962_text.pdf.
M6's verdict flips colitur -> missalemeum: this is a genuine colitur data
gap, the same shape as M4/M7/M9/M14 (absent from data/ef/sanctoral.sexp and
from lectio's own tridentine-calendar.ini), not a spurious oracle entry.
Not fixed here for the same reason those four are not: it requires
touching lectio's own bootstrap source, out of reach for this task.
expected_rows is unchanged (2) -- the diff shape the allow-list predicate
matches does not depend on which side is judged correct, confirmed by the
suite passing unchanged.
A companion register ruling that convicted romanus (9 August) on the
identical reasoning -- and the register's own sanctoral-audit bucket
counts that followed from it -- are corrected in docs/research/
rules-register.md (gitignored, not part of this commit). Full audit
method, the complete list of dropped commemorations found across the
whole calendarium, and every other ruling checked for the same defect
are in .superpowers/sdd/2026-08-12-colitur-rg16a/transcription-audit-
report.md (also gitignored).
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Six independent, small corrections found during the final review:
- dune (workspace root): the comment said the stanza used "(:standard)"
to preserve dune's default `default` alias target; the stanza actually
spells that out explicitly via (alias_rec install). Comment now matches
the code.
- test_validate.ml's test_easter_extremes asserted `List.length ys = 2`
where an identity check was called for -- the comment already named
1598 and 1666, but nothing confirmed extreme_years() found THOSE two
rather than some other pair with the right cardinality. Now asserts the
identities directly (the project's "cardinality where identity was
required" vacuity flavour, per the review).
- test_oracle.ml and expected-divergences-missalemeum.sexp both claimed
"one entry (M13) is [verdict open]" -- M11 is open too (its own verdict
changed from colitur to open in fix round 1); both now say "two entries
(M11 and M13)".
- expected-divergences-missalemeum.sexp's M2 note attributed `band` to
temporal_ef.ml; `band` is precedence_ef.ml's own function.
- lib/kernel/precedence.mli documented `dropped`/`admit`'s physical-
equality obligation nowhere -- it lived only in one rite's own module
(Rite_ef.Precedence_ef.admit's doc comment), but this signature is what
an author of the next rite actually reads. Added the obligation here,
cross-referencing the EF instance as precedent, not the only source.
- README's opam install line omitted sexplib and ppx_sexp_conv (both in
dune-project's own depends; `dune build` fails without them for a
contributor following the README verbatim) and documented only
`colitur easter`, though `temporal` and `day` both exist and are the
more useful entry points. Fixed both.
No behaviour change: comment/doc/test-assertion corrections only (the
easter-extremes fix strengthens an assertion, it does not change what
passes). Verified byte-identical `colitur day` output across 1583, 1900,
1902, 2008, 2011, 2026, 2038, 9999. 259/259 tests green.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
M11 (the Nativity Octave's 26/27/28 December commemoration) previously
read "No RG paragraph found to support this" with verdict colitur.
Wrong on both counts, found by fix-round-1 review.
RG 67 ("dies infra octavam sunt II classis") and RG 109(c) ("de
diebus infra octavam Nativitatis Domini") both name this
commemoration; the 1962 calendarium's own text confirms it directly,
verbatim, under all three dates: "S. STEPHANI PROTOMARTYRIS, II
classis. / Com. octavae Nativitatis." (same pattern for John the
Evangelist and the Holy Innocents).
It is also a genuine regression the previous commit's own
commemoration-eligibility fix introduced: before that fix, colitur
happened to commemorate the day's generic Christmastide loser
unconditionally, which coincidentally kept presence agreeing with the
oracle even though the reasoning was wrong; the fix correctly stopped
that, but 26-28 December's own temporal candidate falls through to
[ferial_rank]'s Christmastide catch-all, which gives it Class4 -- the
wrong ferial rank for a day RG 67 says is II class. A naive fix would
also get RG 69 backwards: when this window falls on a Sunday, RG 69
names the SUNDAY as observed with a commemoration of THE FEAST ITSELF
(Stephen/John/the Innocents), never the generic octave placeholder --
so this needs a commemoration that attaches to whichever named feast
is observed, not one more competing RG 92-95 candidate the way
everything else in this engine is modelled. Out of this fix round's
own scope.
Verdict changed from colitur to open: neither side is confidently
right without that fix, and the previous "colitur is correct, nothing
to implement" framing was itself the risk under review.
|
|
|
Validation layer 4: an oracle harness against missalemeum (Divinum
Officium data), independent of the lectio bootstrap chain colitur's
own sanctoral data comes from -- the only layer that can catch an
error inherited from that bootstrap, and the only one that can
validate commemorations at all (the lectio differential explicitly
excludes them, per its own header comment).
tools/extract_missalemeum_oracle.py shapes the fixture from lectio's
sources/snapshot.tar.gz outside the test (no JSON library in this
project's frozen deps, same reasoning test_differential.ml's own
fixture already documents). test/fixtures/missalemeum-ef-2026-2027
.txt (730 days, SHA-256 pinned and asserted) + its own .provenance
note record exactly how to regenerate it.
test_oracle.ml compares three axes the oracle actually supports:
rank, colour (SET MEMBERSHIP -- 14 of 730 days carry two colours,
e.g. rose+violet on Gaudete/Laetare, which independently vindicates
this project's own rose reading against lectio's violet-only one,
recorded in the register), and commemoration presence/count. Slug
identity is deliberately out of scope (needs a title->slug mapping,
the data audit's own business, not the automated comparator's).
Of 730 days, 688 matched cleanly outright. The remaining 42 are all
named in data/ef/expected-divergences-missalemeum.sexp (14 cited
entries, M1-M14): most are genuine primary-source-confirmed findings
this task adjudicated and fixed in the two preceding commits (RG 33,
RG 109/111, Holy Thursday's colour); the rest are real, cited,
deferred feature/data gaps (RG 91 entry 27's BVM-Saturday office, RG
110's inseparable Peter/Paul commemoration, four sanctoral entries
missing from lectio's own source) or genuine oracle-side artifacts
-- honestly verdicted against whichever side this task's own
primary-source research actually backs, never defaulted to colitur.
One entry (M13, St Joseph vs the Friday of Passion Week 2027) is
verdict open: adjudicated as unresolved after real search effort,
not guessed past.
The data audit: every sanctoral entry the comparison flagged was
hand-checked against the 1962 calendarium, plus a 20-entry
deterministic random control sample (seed 20260812) drawn
independently of the flagged set. The control sample caught two
entries (benedict, frances-rome) marked Commemoration_only in the
bootstrapped data when the primary calendarium lists them as plain
III-class feasts with their own Office -- traced to lectio's own
source, not fixable here, and reported as a signal (10% of a random
sample) rather than a blanket claim. Coverage recorded honestly in
docs/research/rules-register.md's own three buckets: confirmed by
oracle (200/322), confirmed by hand (23/322, 2 of them wrong),
unverified (115/322) -- the unverified bucket stated explicitly
rather than left implicit.
Harness teeth demonstrated and reverted (not committed): a fixture
rank/colour edit on a previously-clean day fails both the checksum
pin and the no-unexplained-differences assertion independently; an
expected_rows drift on the allow-list fails the citation-count
assertion. Both captured with their exact failure messages, both
reverted before this commit.
|
