@classytic/ca-tax 0.0.5 → 0.0.12

package/CHANGELOG.md

@@ -1,5 +1,243 @@
 # Changelog
 
+## [0.0.12] - 2026-08-30
+
+### Fixed
+
+**AT1 Schedule 6's "missing credit line" was not actually a gap — traced to a
+confirmed answer and the incorrect `issues` entry removed.** 0.0.11 shipped
+Schedule 6 (Royalty Tax Credit) with an `issues` entry stating no formula
+existed in the transcribed spec range to compute a final claimable credit
+dollar amount. Tracing the AT1 jacket's own balance-owing formula (spec
+§3.2.3.1, `090 = 080 − (081+082+085+086+087)`) shows it never references a
+Schedule 6 total the way it references Schedule 9's SR&ED credit (081) or
+Schedule 8's political credit (074) — because the **Alberta Royalty Tax
+Credit is administered as an instalment program**, not an annual claimed
+credit. TRA pays or credits ARTC instalments to the corporation during the
+year based on estimated entitlement; the corporation simply reports what it
+received at AT1 jacket line **000082** ("Instalments and other payments and
+ARTC instalments credited to income tax account"), already collected by
+this app's shared "Payments & Instalments" schedule. Schedule 6/7 exist to
+establish the royalty amount, shelter and rate TRA uses for that
+determination — not to produce a number this return deducts. Schedule 6 was
+complete as built; only the messaging (an engine `issues` entry, a module
+docstring, and the UI schedule hint) implied an unresolved defect where none
+existed. Corrected all three.
+
+## [0.0.11] - 2026-08-30
+
+### Added
+
+**Nine entirely-unmodeled AT1 schedules — engine, filing builders, server
+composition and UI, all wired end to end.** Before this release the AT1
+engine modeled 12 of the 21 real Alberta schedules; Schedules 3, 4, 5, 6, 7,
+8, 9, 11 and 15 existed only as bottom-line jacket passthroughs with zero
+supporting-schedule math. Each is now a real, tested module reachable from a
+preparer's form through to a filed Net File payload:
+
+- **Schedule 3** — Other Tax Deductions and Credits: three independent
+  non-refundable credit continuities (Investor Tax Credit, Capital Investment
+  Tax Credit, Agri-Processing Investment Tax Credit) sharing one ceiling
+  (Maximum Allowable Deduction). CITC is gated behind ITC's carryforward;
+  APITC's three earliest vintages each carry their own percentage cap
+  (20%/30%/50%), the rest uncapped.
+- **Schedule 4** — Foreign Investment Income Tax Credit: one row per country,
+  mirroring federal Schedule 21's own occurrences.
+- **Schedule 5** — Royalty Tax Deduction: the Crown Royalty Tax Deduction
+  pool (unsuccessored) plus SSPI/FSPI successored pools, each with its own
+  mandatory (not discretionary) claim formula.
+- **Schedule 6** — Royalty Tax Credit. **Known gap, not guessed at:** the
+  specification's own field map stops at line 006034 — no line anywhere
+  multiplies rate × min(royalty, shelter) into a claimable credit dollar
+  amount. Every other line files; the credit itself is absent with an
+  `issues` entry until an authoritative source resolves it.
+- **Schedule 7** — Royalty Tax Credit/Deduction Supplemental Information:
+  Crown Payment Information, Partnership Income Tax Information and
+  Adjustments to ACR — the disclosure schedule Schedules 5 and 6 both depend
+  on (per the spec, omitting it disallows the credit entitlement outright).
+- **Schedule 8** — Political Contributions Tax Credit: per-contribution
+  receipts plus the three-rate-period tiered credit formula (to-2003,
+  from-2004, and the 2003/2004 straddling tax year).
+- **Schedule 9** — SR&ED Tax Credit (Alberta's own credit, distinct from
+  Schedule 16's federal-mirroring deduction pool), including the page-3
+  associated-group expenditure-limit allocation.
+- **Schedule 11** — Manufacturing and Processing Profits Deduction: a
+  historical schedule (tax years beginning before 2001-04-01 only), gated on
+  both the date and the 10%-of-gross-revenue test.
+- **Schedule 15** — Resource Related Deductions: eight resource-expense
+  continuity pools, each regular + successor (EDA, CMEDB, CEE, CDE, CCOGPE,
+  FEDE, SFEDE, CFRE — the last two per-country). No resource-allowance pool
+  exists on this schedule; it was repealed federally for post-1989 years.
+
+No PDF exists for Schedules 3, 4, 6, 7, 8, 9, 11 or 15 under
+`research/sources/tra-forms/pdf/` — every field number was transcribed
+directly from the TRA specification text (Chapter 3, 2025.2) and cross-
+checked against the engine's own filing-builder tests. Two small,
+purpose-built `FormDefinition`s (`AT1_SCHEDULE_3`, `AT1_SCHEDULE_15`) were
+hand-authored and registered in `src/forms/registry.ts` so every line number
+the return editor shows a preparer is checked, not just asserted — matching
+this package's existing "hand-authored from the specification, no fillable
+form to extract" convention (see `schedule12.ts`).
+
+**Genuine spec ambiguities flagged with an `issues[]` entry rather than
+guessed at, across the nine schedules:** Schedule 3's undocumented cross-
+vintage APITC application order when shared room can't cover every vintage
+(oldest-vintage-first is a documented engine default); Schedule 5's line 025
+formula literally omitting the reimbursements term line 013 includes (kept
+literal, not "fixed"); Schedule 6's missing credit line (above); Schedule
+7's line 061, which has no defining row in its own spec block but is cited
+by Schedule 5's field definition; Schedule 9's line 009031, missing from the
+spec's own field list, derived from the surrounding add/deduct lines with an
+`issues` flag whenever the derived figure is used instead of a direct entry;
+Schedule 11's line 011001, whose caption cites a Schedule 12 line that
+doesn't exist anywhere else in the specification (the computable business-
+rule formula was followed, the discrepancy documented); Schedule 15's
+CCOGPE↔CDE negative-pool routing ambiguity (four lines with an undocumented
+designation gate) and CFRE's "global foreign resource limit" (referenced by
+the claim-cap formula, never defined — treated as nil with a flag rather
+than invented).
+
+`AT1_SCHEDULES_WITH_BUILDERS` now lists 21 schedule codes (was 12).
+`AlbertaReturnInput.schedules` gained nine new optional keys
+(`otherDeductionsCredits`, `foreignInvestmentTaxCredit`,
+`royaltyTaxDeduction`, `royaltyTaxCredit`, `royaltySupplemental`,
+`politicalContributions`, `sredTaxCredit`, `manufacturingProcessing`,
+`resourceDeductions`) alongside the existing eleven.
+
+**Fixed:** a real barrel-export collision discovered mid-build —
+`Schedule4Input`/`Schedule4Result`/`computeSchedule4` and (more seriously)
+`Schedule8Input`/`Schedule8Result`/`computeSchedule8` silently collided with
+unrelated federal T2 exports of the identical names (`schedule4-losses.js`,
+`schedule8.js`, the CCA module) under ESM/TS star-export rules — an explicit
+local export always wins over a same-named star re-export, with no build
+error. Verified live: calling the unprefixed `computeSchedule8` actually
+invoked the federal CCA function and threw `entries is not iterable`. Both
+AT1 pairs are now exported under `Alberta`-prefixed names
+(`computeAlbertaSchedule4`/`AlbertaSchedule4Result`,
+`computeAlbertaSchedule8`/`AlbertaSchedule8Result`).
+
+**Also fixed, found while composing Schedule 17 (Reserves) for this
+release:** `insurancePolicyReserves` and `bankReserves` — the two AT1-only
+reserve kinds with no federal Schedule 13 Part 2 equivalent — are now
+selectable in the UI and carry real Alberta-override wiring
+(`assemble-at1-schedules.ts`'s `scheduleSeventeen`); previously the engine
+supported all 8 reserve kinds but composition only ever passed federal
+figures, so Schedule 17 could never diverge from federal for any kind.
+
+1211 → 1252 tests. 12 → 21 of the 21 real AT1 schedules now modeled (one,
+Schedule 16's SR&ED deduction pool, remains out of scope — no federal T661
+pool module exists to derive it from).
+
+## [0.0.10] - 2026-08-30
+
+### Added
+
+**AT1 Schedule 21's SEVENTH and EIGHTH sections — analysis of losses by year
+of origin (pages 3-4 of the live form) — verified against the actual
+rendered PDF, not just the NetFile spec text.** Two 21-row vintage ledgers:
+
+- `computeNonCapitalLossByYearOfOrigin` (`schedule21-year-of-origin.ts`) —
+  non-capital losses, one row per taxation year (0 = current, 1-20 =
+  preceding — the 20-year expiry limit), six columns per row. Row 0 is
+  fully DERIVED, not asked for twice: the spec states line 157 "must equal
+  021021" (the schedule's own current-year loss) and line 165 "must equal
+  021047" (the pool's total carried back) — both already computed
+  elsewhere. Only the 20 preceding vintages need direct entry, since that
+  history genuinely cannot be derived.
+- `computeOtherLossByYearOfOrigin` — farm / restricted-farm / listed-
+  personal-property, same 21 rows but one raw balance per type per year, no
+  sub-formula. Listed personal property is refused (zeroed, not merely
+  warned) beyond occurrence 7 — it expires after 7 taxation years, not 20.
+
+`schedule21Values` merges both into the same `Schedule Number="021"` block
+as the five pools and the limited-partnership grid — one payload, not four.
+
+**Confirmed out of scope, not silently missing:** the live form's ninth
+section, Continuity of Restricted Interest and Financing Expenses (RIFE,
+page 5, lines 200-250/310-350). An exhaustive search of the entire NetFile
+transmission spec found no `021200`-`021350` field anywhere — RIFE
+continuity exists on the paper form but is not part of the electronic
+filing schema this engine targets.
+
+## [0.0.9] - 2026-08-30
+
+### Added
+
+**AT1 Schedule 21's sixth section — Continuity of Limited Partnership Losses
+(lines 131-141) — was entirely unmodeled (form, engine, and filing all
+missing it) until now.** New `computeLimitedPartnershipLossRow` /
+`computeLimitedPartnershipLosses` (`schedule21-limited-partnership.ts`), one
+row per partnership: `closing = preceding-year balance + wind-up transfer +
+current-year loss − applied`, with `applied` capped at `preceding + wind-up`
+per the spec ("cannot exceed 021133 + 021135"). `schedule21Values` now
+merges these rows into the SAME Schedule 021 payload block as the five
+jurisdiction pools (not a second `Schedule Number="021"` element). The
+column-139 total is exposed on the result for the caller to carry to
+Schedule 12 line 072, per the form's own instruction — that cross-schedule
+wire-up is not done here, same convention as every other Schedule 21 pool's
+Schedule 12 links.
+
+## [0.0.8] - 2026-08-30
+
+### Fixed
+
+**AT1 Schedule 21's loss-continuity formula was missing three terms the real
+form has.** Verified against the spec's own formula (§3.2.3.21): closing
+balance = `(carried forward − expired) + wind-up transfer + current-year loss
+− applied against income − s.80 adjustment − other adjustments − carried
+back`. `computeLossContinuity` previously omitted wind-up transfer, s.80
+adjustment and other adjustments entirely — not just from the UI, from the
+FORMULA. `LossContinuityResult` gains `windUpTransfer`, `section80Adjustment`,
+`otherAdjustments`, and the intermediate `balanceAtBeginningOfYear` (033).
+
+**`schedule21Values` silently dropped every line except opening/current/
+carry-back/closing.** Even where the engine already computed
+`expired`/`appliedCurrentYear` (reused from federal), the filing builder never
+wrote lines 032/035/041/043/045 (and each pool's equivalents) to the XML —
+they existed in memory and never reached TRA. Rewritten to read the pool's
+own line numbers directly off `AT1_SCHEDULE_21_POOLS` (the form definition's
+own table, not a second hand-copied one) and emit every field the pool's
+form entry actually has a box for.
+
+## [0.0.7] - 2026-08-29
+
+### Added
+
+**A genuine NEW Class 13/14 addition now amortises correctly, on both
+federal Schedule 8 and AT1 Schedule 13.** 0.0.6's Class 13/14 support (added
+in 0.0.4) only ever handled an EXISTING opening-UCC pool with no addition —
+correct for that case, but it refused a real new leasehold layer or
+limited-life property outright, even though the full Reg 1100(1)(b)/(c) /
+Schedule III mechanic (`computeClass13`/`computeClass14`) already existed in
+`cca-straight-line.ts` and just wasn't wired into either pipeline.
+
+- `FederalT2Input` gains `class13?: Class13Input` / `class14?: Class14Input`
+  — separate from `ccaClasses`, which keeps its existing-balance-only
+  behaviour. Injects the Schedule 1 deduction (line 403, combined with any
+  declining-balance CCA into one line) same as before.
+- `AlbertaSchedule13Input` gains `class13?: AlbertaClass13Input` /
+  `class14?: AlbertaClass14Input` — federal and Alberta share the same
+  layers/properties (a lease term doesn't change by jurisdiction); only the
+  CLAIM can diverge, same "omit = defaults to federal's actual claim"
+  semantics every other AT1 column already has. Alberta's claim is still
+  capped at the same Schedule III maximum federal's is — a discretionary
+  election can claim less, never more than the layer's own ceiling allows.
+- Verified against a hand-derived example (a $25,000, 7-year leasehold,
+  first year): the 5-year floor and half-year rule correctly cap the year-1
+  claim at $3,571, not the full $25,000 the old degenerate stub would have
+  claimed.
+
+## [0.0.6] - 2026-08-29
+
+### Changed
+
+**Raised the `@classytic/primitives` / `@classytic/tax-core` peer floors.**
+They were still declared at `>=0.9.0` / `>=0.1.0` — stale by many minor
+releases (current: 0.26.1 / 0.3.0) and well below what every actual consumer
+already runs (`apps/server` uses `^0.23.0` / `^0.3.0`). Raised the peer and
+dev floors to `>=0.23.0` / `>=0.3.0`, matching the versions this package is
+actually built and tested against.
+
 ## [0.0.5] - 2026-08-29
 
 ### Added