The audit record · generated, never hand-edited

Conformance record

Every form compared with the CRA Payroll Deductions Online Calculator (PDOC), every disagreement listed with its inputs and both outputs. Generated by tools/conformance from the committed PDOC outputs; latest capture 2026-09-15.

Default arm, t4127

8,905/9,002

98.92% of forms match on all seven lines

rounding_compat: pdoc

8,987/9,002

99.83% with the CPP line in binary64, as inferred from PDOC's outputs

Unexplained

15 forms

One cent each, counted against both rates, listed below

Oracle is CRA PDOC (https://apps.cra-arc.gc.ca/ebci/rhpd/beta/). PDOC version observed on filled vectors: 2026-06-11 (M1); grid queue: not recorded (the capture read no form controls). Capture operator: pdoc-oracle-harness.

Expected amounts are PDOC-only. They are never filled from the engine.

This file is generated by tools/conformance (takehome-conformance). Do not hand-edit rates or counts; change the run and regenerate.

Overall agreement rate (PDOC corpora): 8905/9002.

With rounding_compat: pdoc (CPP line in binary64, as inferred from PDOC's outputs): 8987/9002.

The engine reproduces PDOC's CPP line exactly under binary64 shortest-decimal rounding (the rounding_compat: pdoc arm). The default arm follows T4127 plus two PDOC-matched roundings T4127's text does not specify: K2/K2P credits and Ontario surtax lines are rounded separately. Both rates count every compared form; nothing is exempted. The July 2026 grid was captured from live PDOC (last capture 2026-09-15). See M-004 and finding 003.

The PDOC corpora use TD1 claim codes 0, 1 and E only, so neither rate says anything about claim codes 2 to 10. Those rest on the CRA claim-code tables and an independent oracle (tools/claim-codes/oracle.py, 384 vectors in crates/takehome-core/tests/vectors/claim_codes_2026.json).

Methodology

A disagreement is: correctly-specified identical inputs, engine and PDOC both given exactly what the operator intended, outputs differ. That is a finding about the world. It is published here with the full input, both outputs, the delta, and an explanation: including cases where PDOC differs from a literal reading of T4127.

A disagreement is not: a bug in the harness, a mis-mapped request or result field, or a vector whose input was specified differently on the two sides. That is a defect. Fix it, add a regression test, and it never appears in this record.

The distinction is the document’s credibility. Inflating the disagreement count with our own mapping bugs is less trustworthy, not more honest. Publishing everything is not the same as publishing indiscriminately. The number of disagreements below means genuine remaining deltas, not “every time a test failed.”

Until every defined vector in a corpus is measured, this document states n, how many are measured, and how many are pending. It does not compute an agreement rate on a partial run. A completed corpus states how many match and lists every disagreement.

M-001: Published whole-dollar K (ADR-002)

Annual T3/T4 is discontinuous by at most $1.00 at each bracket threshold, because the CRA publishes the K constants rounded to whole dollars. PDOC exhibits the same discontinuity; we match it deliberately.

The per-threshold jump equals the difference of that bracket's two K rounding residuals. Measured 2026 jumps for **all thirteen bracket tables** (FED plus twelve provinces/territories; both January and July editions; Outside Canada has no provincial K) live in crates/takehome-core/tests/vectors/bracket_discontinuity_2026.json.

Period-tax vs gross is therefore not strictly monotone. The property ontario_calculate_invariants uses the M-001 bound (≤ $0.09 per period across sampled P ≥ 12). The thirteen-jurisdiction property calculate_invariants_all_jurisdictions scales the same residual difference by P, adds provincial surtax amplification and the Outside Canada 48% T1 surtax, and allows two cents of period rounding. #[ignore] is never a resolution for those tests.

Decision: docs/ADR-002-published-constants.md.

M-002: PDOC K2 maximum in the reaching period

T4127 Chapter 3 tells the calculator to use the maximum base CPP contribution “in that pay period”: the period year-to-date first reaches the annual CPP maximum. PDOC does not. Its federal CPP credit base is max(P × C × ratio, D × ratio) capped at base_max, which on the measured vector is D × ratio = 3494.12 rather than base_max = 3519.45.

The period-tax difference is 13¢ federal and 5¢ provincial on on-biweekly-midyear-k2-max. That is a fact about PDOC, not an engine defect: the default k2_method is pdoc_observed because customers reconcile against the oracle. The document’s reading is selectable as t4127_literal.

Full write-up (vector, both outputs, delta, inversion, reproduction): docs/findings/001-k2-maximum-in-reaching-period.md.

PDOC mapping notes: employee.federal_tax is T4127 [(T1)/P]+L (PDOC Federal + Additional). employee.total_tax is the sum of the two separately rounded federal and provincial lines a user can add up on screen; breakdown.T is T4127 Step 6 round((T1+T2)/P)+L and may differ by one cent. Claim code 1 must be entered via PDOC Claim codes, not a fixed TD1 dollar amount: fixed $16,452 does not phase out BPAF. employee.total_deductions includes union dues.

M-004: PDOC's CPP line matches binary64 (closed 2026-10-22)

Inferred from PDOC's outputs (its code is not public): PDOC's period CPP contribution matches (PI - exemption) x 0.0595 evaluated in IEEE-754 binary64, then rounded half-up from the shortest decimal of that value. On an exact decimal half-cent (150.00 x 0.0595 = 8.925) binary64 holds 8.924999999999999, so PDOC shows 8.92; T4127 decimal gives 8.93. All 82 CPP-line disagreements in the July queue are this case; the rule changes none of the 8,900 CPP lines that already agreed. The default arm keeps decimal arithmetic for C. rounding_compat: pdoc reproduces PDOC with an integer-only binary64 emulator (no floats in the money path). Finding: docs/findings/003-pdoc-cpp-binary64-and-split-credits.md. Decision: docs/ADR-007-pdoc-binary64-cpp.md.

M-003: PDOC midpoint direction (mostly closed 2026-10-22)

Finding 003 closed 149 of the 164 one-cent disagreements with three named rules: K2/K2P as two separately rounded credits, Ontario V1 as two separately rounded surtax lines (both in the default arm) and M-004 above. The original record follows for history.

M-003 (history): PDOC midpoint direction

Live cent delta: T4127 half-up at an exact midpoint goes up; PDOC’s line is sometimes 1¢ lower. Counts against the agreement rate: not exempted as methodology. Not Alberta-specific; AB is dense because 8% flat on round grosses manufactures halves. Observed down-classes: AB provincial T2/P, AB CPP, NL 4333.33 P=12, NL 1718.39 P=26 (neighbour 1718.40 goes up), NU 2536.41 P=22. Cross-jurisdiction probes: 47/50 discriminating half-cent forms match engine (up). Condition that selects down vs up is unnamed. Finished July queue: 164 one-cent disagreements across thirteen jurisdictions (AB 40, ON 18, BC 14, NU 14, NS 12, MB 11, YT 11, NL 10, PE 8, SK 8, NB 6, NT 6, OutsideCanada 6); 780 forms the salary UI accepted as queue items but PDOC step-2 would not advance. Those 780 are named, not pending, and not in the rate.

Product: ADR-003: rounding_compat (t4127 default; pdoc implemented 2026-10-22 per ADR-007 for the CPP line).

Full write-up: docs/findings/002-pdoc-midpoint-direction.md.

M4 named classes (not folded into the Option 1 rate)

PDOC evidence publication (ruling a)

JSON cache records are published as the GitHub release archive takehome-conformance-corpus-2026.1.tar.zst, not in git. pdoc-identity.json and screenshot-hashes.json stay in the repo so the catalog digest is checkable without a 38 MB clone. PNG screenshots stay local (data/pdoc-screenshots/, gitignored). This document records the catalog digest, the archive SHA-256, and the twenty M1 screenshot hashes. Publication tests extract the archive and break the catalog on purpose.

Sampling design

Published PDOC queue: every distinct July live-edition boundary form the salary UI can enter: ten of fourteen legal P, both claim codes, fingerprint-deduped. The earlier 15,264 figure assumed all fourteen P; PDOC cannot capture annual / semi-annual / quarterly / hourly (1, 2, 4, 2000). Those forms, plus $0 and sub-dollar gross, are the named uncapturable class below: validated by invariants and the differential oracle (same as the interior ladder), not by silently shortening a queue that still claims 15,264.

The spec’s “250,000 cases” target is a case *list*, not a PDOC queue. At 3 seconds per uncached form, that queue does not finish. The adopted split is: **every capturable unique live-edition boundary form goes to PDOC; the interior log-ladder and the uncapturable class do not.**

Field Value
grid_version 2026.1
grid cells 1263636
unique July boundary forms, all 14 legal P (superseded) 15264
PDOC queue (10 capturable P, distinct after dedup) 9762
unique July uncapturable forms 5502
wall clock at 3 s / uncached form 8h 8m
overnight 8 h capacity 9600
log-ladder cells (invariants only) 1123344
DateBeforeCoverage cells (engine error) 140354

Grid cells collapse heavily onto PDOC fingerprints; the finished capture count will be at most the distinct-form figure above (smoke already showed cache hits from overlapping forms).

Run conditions

Field Value
grid_version 2026.1
engine_build_sha256 1704657a70ddd0a693d72dc9355f46732bafeb4a70228c30c2a8afd3c0ee0aec
rule_set_versions 2026-01-01, 2026-07-01, 2027-01-01
rule_set_sha256 a398e70a9c904d5dc21ee55fe104383c9814cb1c7d131164da668ef6a96017e5
PDOC identity 2026-06-11 (M1); grid queue: not recorded (the capture read no form controls)
browser chromium 153.0.8010.12
harness pdoc-oracle-harness
latest PDOC retrieved_at 2026-09-15T03:41:47.438Z

Cases by oracle class

Each grid cell is named with the oracle allowed to validate it. Boundary cells that share a PDOC-visible fingerprint collapse to one form; the queue size is the unique-form count above, not the cell count.

Jurisdiction P Rule set Oracle class Cells
AB 1 2026-01-01 invariants 2328
AB 1 2026-01-01 uncapturable 396
AB 1 2026-07-01 invariants 2328
AB 1 2026-07-01 uncapturable 396
AB 1 2027-01-01 invariants 1552
AB 1 2027-01-01 uncapturable 264
AB 1 none engine_error 908
AB 2 2026-01-01 invariants 2328
AB 2 2026-01-01 uncapturable 324
AB 2 2026-07-01 invariants 2328
AB 2 2026-07-01 uncapturable 324
AB 2 2027-01-01 invariants 1552
AB 2 2027-01-01 uncapturable 216
AB 2 none engine_error 884
AB 4 2026-01-01 invariants 2322
AB 4 2026-01-01 uncapturable 246
AB 4 2026-07-01 invariants 2322
AB 4 2026-07-01 uncapturable 246
AB 4 2027-01-01 invariants 1548
AB 4 2027-01-01 uncapturable 164
AB 4 none engine_error 856
AB 10 2026-01-01 pdoc 234
AB 10 2026-01-01 invariants 2256
AB 10 2026-01-01 uncapturable 12
… … … … 1640 more slices in conformance.json

Full slice table: `conformance.json` census.

Corpora

m1-ontario-2026-01

Field Value
oracle class pdoc
defined 20
measured 20
pending 0
exact matches 20
agreement rate, rounding_compat: pdoc 20/20
agreement rate 20/20

pdoc-bonus-2026

Field Value
oracle class pdoc
defined 20
measured 20
pending 0
exact matches 19
agreement rate, rounding_compat: pdoc 19/20
agreement rate 19/20

year-projection-2026

Field Value
oracle class invariants
defined 8
measured 8
pending 0
exact matches 8
agreement rate *not a PDOC corpus; year-projection invariants (API tests 22-29), not per-cell PDOC*

t4127-option2-2026

Field Value
oracle class t4127_worked_examples
defined 3
measured 3
pending 0
exact matches 3
agreement rate *not a PDOC corpus; T4127 Chapter 5 worked examples + invariants. Weaker evidence than PDOC. Does not inherit the Option 1 PDOC agreement rate.*

grid-2026.1-pdoc-boundary

Field Value
oracle class pdoc
defined 9762
measured 8982
pending 0
exact matches 8885
agreement rate, rounding_compat: pdoc 8967/8982
agreement rate 8885/8982

grid-2026.1-uncapturable-boundary

Field Value
oracle class uncapturable
defined 5502
measured 0
pending 5502
exact matches 0
agreement rate *not a PDOC corpus; invariants + differential oracle (§16.2), not per-cell PDOC*

grid-2026.1-interior-invariants

Field Value
oracle class invariants
defined 1126620
measured 0
pending 1126620
exact matches 0
agreement rate *not a PDOC corpus; invariants + differential oracle (§16.2), not per-cell PDOC*

Disagreements

id provincial_tax federal_tax cpp net_pay
on-biweekly-2000-bonus-1000 - -0.01 - 0.01
AB-P10-500.00-F0-P0 - - 0.01 -0.01
AB-P10-500.00-F1-P1 - - 0.01 -0.01
AB-P10-7460.00-F0-P0 - - 0.01 -0.01
AB-P10-7460.00-F1-P1 - - 0.01 -0.01
AB-P10-8500.00-F0-P0 - - 0.01 -0.01
AB-P10-8500.00-F1-P1 - - 0.01 -0.01
BC-P10-500.00-F0-P0 - - 0.01 -0.01
BC-P10-500.00-F1-P1 - - 0.01 -0.01
BC-P10-7460.00-F0-P0 - - 0.01 -0.01
BC-P10-7460.00-F1-P1 - - 0.01 -0.01
BC-P10-8500.00-F0-P0 - - 0.01 -0.01
BC-P10-8500.00-F1-P1 - - 0.01 -0.01
BC-P240-106.55-F1-P1 - -0.01 - 0.01
MB-P10-40000.00-F0-P0 - - 0.01 -0.01
MB-P10-40000.00-F1-P1 - - 0.01 -0.01
MB-P10-500.00-F0-P0 - - 0.01 -0.01
MB-P10-500.00-F1-P1 - - 0.01 -0.01
MB-P10-7460.00-F0-P0 - - 0.01 -0.01
MB-P10-7460.00-F1-P1 - - 0.01 -0.01
MB-P10-8500.00-F0-P0 - - 0.01 -0.01
MB-P10-8500.00-F1-P1 - - 0.01 -0.01
MB-P24-1958.33-F0-P0 - 0.01 - -0.01
MB-P24-1958.33-F1-P1 - 0.01 - -0.01
NB-P10-500.00-F0-P0 - - 0.01 -0.01
NB-P10-500.00-F1-P1 - - 0.01 -0.01
NB-P10-7460.00-F0-P0 - - 0.01 -0.01
NB-P10-7460.00-F1-P1 -0.01 - 0.01 -
NB-P10-8500.00-F0-P0 - - 0.01 -0.01
NB-P10-8500.00-F1-P1 - - 0.01 -0.01
NL-P10-500.00-F0-P0 - - 0.01 -0.01
NL-P10-500.00-F1-P1 - - 0.01 -0.01
NL-P10-7460.00-F0-P0 - - 0.01 -0.01
NL-P10-7460.00-F1-P1 - - 0.01 -0.01
NL-P10-8500.00-F0-P0 - - 0.01 -0.01
NL-P10-8500.00-F1-P1 - - 0.01 -0.01
NL-P53-842.97-F0-P0 - -0.01 - 0.01
NL-P240-930.59-F1-P1 -0.01 - - 0.01
NS-P10-500.00-F0-P0 - - 0.01 -0.01
NS-P10-500.00-F1-P1 - - 0.01 -0.01
NS-P10-7460.00-F0-P0 - - 0.01 -0.01
NS-P10-7460.00-F1-P1 - - 0.01 -0.01
NS-P10-8500.00-F0-P0 - - 0.01 -0.01
NS-P10-8500.00-F1-P1 - - 0.01 -0.01
NS-P22-2817.78-F0-P0 0.01 - - -0.01
NS-P240-258.29-F1-P1 -0.01 - - 0.01
NT-P10-500.00-F0-P0 - - 0.01 -0.01
NT-P10-500.00-F1-P1 - - 0.01 -0.01
NT-P10-7460.00-F0-P0 - - 0.01 -0.01
NT-P10-7460.00-F1-P1 - - 0.01 -0.01
NT-P10-8500.00-F0-P0 - - 0.01 -0.01
NT-P10-8500.00-F1-P1 - - 0.01 -0.01
NU-P10-500.00-F0-P0 - - 0.01 -0.01
NU-P10-500.00-F1-P1 - - 0.01 -0.01
NU-P10-7460.00-F0-P0 - - 0.01 -0.01
NU-P10-7460.00-F1-P1 - - 0.01 -0.01
NU-P10-8500.00-F0-P0 - - 0.01 -0.01
NU-P10-8500.00-F1-P1 - - 0.01 -0.01
NU-P240-1077.00-F0-P0 -0.01 - - 0.01
NU-P240-310.82-F1-P1 -0.01 - - 0.01
ON-P10-500.00-F0-P0 - - 0.01 -0.01
ON-P10-500.00-F1-P1 - - 0.01 -0.01
ON-P10-5852.29-F1-P1 -0.01 - - 0.01
ON-P10-7460.00-F0-P0 - - 0.01 -0.01
ON-P10-7460.00-F1-P1 - - 0.01 -0.01
ON-P10-8500.00-F0-P0 - - 0.01 -0.01
ON-P10-8500.00-F1-P1 - - 0.01 -0.01
ON-P13-2769.24-F0-P0 - 0.01 - -0.01
ON-P13-447.54-F0-P0 - 0.01 - -0.01
ON-P13-447.55-F0-P0 - 0.01 - -0.01
OutsideCanada-P10-500.00-F0-P0 - - 0.01 -0.01
OutsideCanada-P10-500.00-F1-P1 - - 0.01 -0.01
OutsideCanada-P10-7460.00-F0-P0 - - 0.01 -0.01
OutsideCanada-P10-7460.00-F1-P1 - - 0.01 -0.01
OutsideCanada-P10-8500.00-F0-P0 - - 0.01 -0.01
OutsideCanada-P10-8500.00-F1-P1 - - 0.01 -0.01
PE-P10-500.00-F0-P0 - - 0.01 -0.01
PE-P10-500.00-F1-P1 - - 0.01 -0.01
PE-P10-7460.00-F0-P0 - - 0.01 -0.01
PE-P10-7460.00-F1-P1 - - 0.01 -0.01
PE-P10-8500.00-F0-P0 - - 0.01 -0.01
PE-P10-8500.00-F1-P1 - - 0.01 -0.01
PE-P240-445.37-F1-P1 -0.01 - - 0.01
SK-P10-500.00-F0-P0 - - 0.01 -0.01
SK-P10-500.00-F1-P1 - - 0.01 -0.01
SK-P10-7460.00-F0-P0 - - 0.01 -0.01
SK-P10-7460.00-F1-P1 - - 0.01 -0.01
SK-P10-8500.00-F0-P0 - - 0.01 -0.01
SK-P10-8500.00-F1-P1 - - 0.01 -0.01
SK-P12-4544.33-F1-P1 0.01 - - -0.01
YT-P10-500.00-F0-P0 - - 0.01 -0.01
YT-P10-500.00-F1-P1 - - 0.01 -0.01
YT-P10-50000.00-F0-P0 - - 0.01 -0.01
YT-P10-50000.00-F1-P1 - - 0.01 -0.01
YT-P10-7460.00-F0-P0 - - 0.01 -0.01
YT-P10-7460.00-F1-P1 - - 0.01 -0.01
YT-P10-8500.00-F0-P0 - - 0.01 -0.01
YT-P10-8500.00-F1-P1 - - 0.01 -0.01

Full request / engine / PDOC maps: `conformance.json` corpora[].disagreements.

on-biweekly-2000-bonus-1000

engine and PDOC differ on identically specified bonus inputs

AB-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

AB-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

AB-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

AB-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

AB-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

AB-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

BC-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

BC-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

BC-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

BC-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

BC-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

BC-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

BC-P240-106.55-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

MB-P10-40000.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P10-40000.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

MB-P24-1958.33-F0-P0

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

MB-P24-1958.33-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

NB-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NB-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NB-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NB-P10-7460.00-F1-P1

M-004 (finding 003): the CPP line differs by one binary64 cent and the tax line follows through K2. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NB-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NB-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NL-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NL-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NL-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NL-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NL-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NL-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NL-P53-842.97-F0-P0

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

NL-P240-930.59-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

NS-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NS-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NS-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NS-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NS-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NS-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NS-P22-2817.78-F0-P0

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

NS-P240-258.29-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

NT-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NT-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NT-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NT-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NT-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NT-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NU-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NU-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NU-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NU-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NU-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NU-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

NU-P240-1077.00-F0-P0

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

NU-P240-310.82-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

ON-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

ON-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

ON-P10-5852.29-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

ON-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

ON-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

ON-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

ON-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

ON-P13-2769.24-F0-P0

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

ON-P13-447.54-F0-P0

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

ON-P13-447.55-F0-P0

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

OutsideCanada-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

OutsideCanada-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

OutsideCanada-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

OutsideCanada-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

OutsideCanada-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

OutsideCanada-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

PE-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

PE-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

PE-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

PE-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

PE-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

PE-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

PE-P240-445.37-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

SK-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

SK-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

SK-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

SK-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

SK-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

SK-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

SK-P12-4544.33-F1-P1

M-003 residual (finding 003): one-cent period tax line. The engine's T/P sits within a tenth of a cent of a half-cent; no published rounding rule we tested reproduces PDOC here (see the hypothesis table). Counts in both rates.

YT-P10-500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

YT-P10-500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

YT-P10-50000.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

YT-P10-50000.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

YT-P10-7460.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

YT-P10-7460.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

YT-P10-8500.00-F0-P0

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

YT-P10-8500.00-F1-P1

M-004 (finding 003): PDOC's period CPP line matches IEEE-754 binary64 arithmetic (inferred from its outputs). The exact decimal product is a half-cent that binary64 holds below the half, so PDOC shows one cent less than T4127 half-up. Matches under rounding_compat: pdoc. Counts in the default-arm rate.

Closed: D-007 biweekly mid-year K2

Vector: on-biweekly-midyear-k2-max (matches under pdoc_observed). Published as M-002. PDOC does not force base_max in the reaching period. That is a fact about PDOC, stated with inversion in findings/001.

Closed 2026-10-22: most of M-003 (finding 003)

Of the 164 one-cent July disagreements, 82 were the CPP line and 82 were tax lines. Three named rules close all but 15 of them, with no free parameter:

  1. K2 and K2P are two credits (CPP and EI), each rounded to the cent, as the T1 return computes them. T4127 writes two products but does not say where to round; this matches PDOC. Default arm.
  2. Ontario V1 is two surtax lines (20% and 36%), each rounded. Default arm.
  3. M-004: PDOC's C matches IEEE-754 binary64 (inferred from its outputs): binary64 arithmetic, then the shortest decimal of that value rounded half-up. rounding_compat: pdoc only; the default arm keeps decimal arithmetic for C (8.925 rounds to 8.93; PDOC shows 8.92).

The rules were chosen by inspecting the full corpus; 15 cases remain unexplained. The integer binary64 emulator is differential-tested against hardware floating point on 10 million CPP inputs. findings/003.

Open: M-003 residual

The disagreements left under rounding_compat: pdoc are one-cent period tax lines whose engine T/P sits within a tenth of a cent of a half-cent. No published rounding rule we tested reproduces them; they count in both rates.

Cases where PDOC differs from T4127 decimal

M-004: on an exact decimal half-cent, PDOC's CPP line shows one cent less (8.925 shown as 8.92), consistent with binary64 arithmetic. T4127 decimal arithmetic rounds up. Both are measured and published; PDOC's result is selectable as rounding_compat: pdoc. M-002 remains a selectable k2_method difference.

Coverage gaps

PDOC evidence publication

Ruling (a): JSON cache records are published as a release archive, not in git. PNG screenshots stay local. Their SHA-256 values are recorded here so a missing file cannot stay green only because the report never looked at it.

Field Value
ruling records-only
records in git no (data/pdoc-cache/records/ stays local; 9010 forms)
archive takehome-conformance-corpus-2026.1.tar.zst
archive SHA-256 00a68d0bef8835f8271039981c22bcaca54ff5f0786c33fd3c5b346b27a0023f
download https://github.com/GautamTalksDev/takehome/releases/download/conformance-corpus-2026.1/takehome-conformance-corpus-2026.1.tar.zst
screenshots published no (data/pdoc-screenshots/, gitignored)
screenshot catalog `data/pdoc-cache/screenshot-hashes.json`
catalog digest (SHA-256 of sorted key<space>sha256 lines) 554a54278727bb460e7f2710cde2b6e31b750ea57c4a4ee6735eb2e4d02a71ea

After download, a stranger can check the claim:

echo 00a68d0bef8835f8271039981c22bcaca54ff5f0786c33fd3c5b346b27a0023f  takehome-conformance-corpus-2026.1.tar.zst | sha256sum -c
tar --zstd -xf takehome-conformance-corpus-2026.1.tar.zst && python3 -c "import json,hashlib,pathlib;h={d['key']:d['screenshotSha256'] for d in (json.loads(p.read_text()) for p in pathlib.Path('records').glob('*.json'))};digest=hashlib.sha256(''.join(f'{k} {h[k]}\n' for k in sorted(h)).encode()).hexdigest();assert digest=='554a54278727bb460e7f2710cde2b6e31b750ea57c4a4ee6735eb2e4d02a71ea'"

M1 screenshot hashes (local PNGs)

id sha256
on-weekly-600-cc1 6e90c7ca206a1f07632715f82276c2d0134cc6f7c50e142eae6ee52bf717e3e5
on-weekly-1000-cc1 ecac47e3146e41ff1fbd4a56318a63c8db2389fa50a37d50203c32756dfde806
on-weekly-2000-cc1 a1b566e315ffe1e519a59648a2e9646a1b2064b8976a1806d4591bd4f4ddef47
on-weekly-above-ympe 3e6cbfb0f0c62c4cca7c6634ba4b0f3d3a865f951f5a21c46a1956365b9e9c92
on-weekly-above-yampe e5396f02628950b636f91700dd592b8890d1b9ce5258ce4f76f8d78bcbaf616f
on-weekly-bpaf-phaseout c6b0c671c880d3ad5312fc82a4b724ce6c70b114cf66e517147776781367be31
on-weekly-surtax-tier1 39a8d38e824922be04b59fc089067bd19291c94ebd539da244ba798e3089efcf
on-weekly-surtax-tier2 36badeb15aee410cad5d7be1b14a49b83b11d156e07abfa1170cd1b96d8a7b15
on-v2-tier0-at-or-below-20k 5703bdb1253413de765e87c55b64fe1a3b018d6e5b562bbbccff7c29efae86be
on-v2-tier1-20k-36k 9fe68bf570ce11783b7c20b1a9f4dd8d993ce8f364d96a00d1aa04bc58be2c48
on-v2-tier2-36k-48k d60354ddb5f7fa18fe148345166b53aa890677765e40fc0c4124f4d031371529
on-v2-tier3-48k-72k 5298a5310dbc198164097bf781eb5a2a6381cf5238cdaa82f85a46383d68e712
on-v2-tier4-72k-200k cfa9c8f17f957767de88433d8550405e76389e13b29c745be6085a52b664732c
on-v2-tier5-above-200k 41f95b80b1a7b7fb3c615863768563a532a9cf47ff137a5c740fd3f5863625c2
on-weekly-claim-code-e 4638100cbb54d573c8e86721c25fc7d0a0dc51e0e4f247122b04e63b3671df36
on-weekly-claim-code-0 584685a947f4b9f6c39e341971d01d8449f972e74b2576fd6dfa2fd0ab97d2b2
on-weekly-direct-tc-tcp e4237a6dfceca01aad822854fac07dbe09b7b38f04d40c55610cd308b4075920
on-monthly-1000-cc1 82a6d05b3d3cbf8637a08eb0a0a9c3a1e9f3935deaea491b5cca279ad5ce064e
on-biweekly-midyear-k2-max 7ffb4f821d664bba20b2c09b05903520505344f10c8dab61bca5f772f155e021
on-weekly-l-and-u1 af83d873b3c5099adde12dfe9bec2aed0fac85e1f16153e30aff786eb1b25d77

M1 measured results (complete corpus)

Ontario 2026-01, twenty vectors, operator manual, PDOC 2026-06-11. All twenty match. See crates/takehome-core/tests/vectors/pdoc_ontario_2026_01.json.

M2 exit checklist

Criterion State
Ingest round-trips to hand-authored Ontario and federal files yes (test 6)
Thirteen jurisdictions, both 2026 editions, test 28 and 28b yes
July delta overlay; ten unchanged jurisdictions byte-identical yes
BC/NL/PE OptionScoped proration on both options yes
2026-06-30 / 2026-07-01 boundary pair BC, NL, PE, ON yes
Quebec typed unsupported, never federal-only yes
Grid deterministic, versioned, boundary-present, even yes (2026.1)
Oracle sampling design decided, public, finishable yes (this section)
Conformance report with the real number and every disagreement this file
Findings 001 (K2) and M-001 discontinuity written up yes
Tagged v0.3.0-m2 after this report is committed

M4 exit checklist

Criterion State
Batch endpoint, 25 items (the Workers Free cap), partial success, metering yes
Year projection CPP/EI caps on the annual maximum yes
CRA bonus worked example TB = 323.54 yes
Option 2 BC/NL/PE proration (PDOC does not expose Option 2) yes
Rule-change detection; signed retrying webhooks yes
2027 preview rule set, proposed, warning asserted yes
Conformance re-run, no PDOC regression, new classes named this file
pdoc-bonus-2026 captured, own class, not in the overall Option 1 rate yes
rounding_compat: pdoc implemented (CPP binary64, finding 003) yes (2026-10-22)
Tagged v0.5.0-m4 after this report is committed
Five-minute classmate on production open (docs/FIVE-MINUTE-TEST.md)
npm / PyPI takehome-ca open (no publish tokens here)

M0 arithmetic canaries