Integrate 15 reviewed primitive and Taproot improvements - #246
Merged
Merged
Conversation
Resolve the seven conflicted files by keeping main's content and adding the
ternary hash-path entries:
- knowledge/catalog.json: take main's schema 1.1 header; keep all 101 main
records unchanged and add commitment/ternary-hash-path-integer (102 records,
257 configurations).
- knowledge/negative-results/index.md: rename the PR's colliding NR-043
("Ternary mixed-hash paths lose to four-way integer paths") to NR-072 and
append it after main's NR-065; main's NR-043 is untouched. Update its
stale 924-byte figure to the current 947 bytes.
- knowledge/open-problems.md: the PR's OP-020 collided with main's OP-020
(bound-start hash paths); renumber it OP-030 (OP-027 is used by two other
open PRs) and update the catalog reference.
- knowledge/comparisons/commitments.md, knowledge/index.md: keep main's rows,
paragraphs and review date; add the ternary row (947/63/24) and a paragraph
with its 509-byte gap, data/hint items and execution class.
- src/commitments/README.md, tests/primitive_metrics.rs: follow main's
per-primitive layout. Move the module to src/commitments/ternary_hash_path/
with its own README (template sections, explicit 0 hint items), and add a
ternary_hash_path_metrics() group chained into metrics() and checked by the
existing named ternary_hash_path_metrics_are_current test. The stack metric
now uses the strict executor; values are unchanged (947/63/24).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add rejects_first_out_of_range_value_at_widths_1_and_31_before_overflow. For widths 1 and 31 it checks a valid 2^width-1 control and requires 2^width to be rejected with ExecError::Verify from the pre-final-step width check. At width 31 a bound applied only after the last multiply/add is rejected by ScriptNum overflow instead, which the existing all-width test cannot distinguish. The pinned interpreter (a09e87af) now counts every tapscript instruction position in opcode_count, so the benchmark's executed_opcodes=919 was a static position count, not an executed-opcode total, and the documented value 0 was stale. Report static instructions (919), static non-push opcodes (794), the interpreter position count and executed_opcodes=unavailable, and correct the README, knowledge page and research note. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…e ternary path Second integration of origin/main (3d1001f, after #244). Conflicts: - src/commitments/mod.rs: keep main's tapbranch module/exports and the ternary_hash_path module/exports (alphabetical order). - tests/primitive_metrics.rs: keep main's tapbranch imports and the ternary imports; all 134 main tests/931 metric keys remain, plus the ternary test and its three keys. - knowledge/comparisons/commitments.md: keep main's TapBranch u4 row and the ternary row. - knowledge/negative-results/index.md: keep main's NR-066 and append the ternary entry as NR-072 after it (68 unique NR headings). Catalog auto-merged: 117 main records unchanged plus the ternary record (118 records, 284 configurations). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Resolve conflicts with main's integrated u4 siblings (threshold, equality, trichotomy, clamp, MSB, and others) by keeping both sides: - knowledge/catalog.json: main's 117 records unchanged plus the PR's arithmetic/u4-count record (118 records, 284 configurations). - tests/primitive_metrics.rs: main's 134 metric tests plus u4_symbol_count_metrics_are_current (135). - src/arithmetic/u4/mod.rs, src/arithmetic/u4/README.md, knowledge/comparisons/arithmetic.md, knowledge/primitives/index.md: union of module declarations, API bounds, and table rows. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Resolve conflicts against origin/main 3d1001f by keeping both sides: - knowledge/comparisons/arithmetic.md, knowledge/primitives/index.md, src/arithmetic/u32/README.md: keep main's new rows/bullets/paragraphs and the constant XOR row/bullet/paragraph. - src/arithmetic/u32/mod.rs: keep xnor_constant/zero/zero_byte_mask from main and add xor_constant. - tests/primitive_metrics.rs: take main and re-add u32_xor_constant_metrics_are_current (134 metric tests; main's 133 plus 1). - knowledge/catalog.json merged cleanly: 118 records/284 configurations (main 117/283 plus arithmetic/u32-xor-constant/checked-constant). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Resolve conflicts in knowledge/catalog.json, knowledge/comparisons/arithmetic.md, knowledge/primitives/index.md, src/arithmetic/u4/README.md, src/arithmetic/u4/mod.rs and tests/primitive_metrics.rs by keeping every main-side record, row, module and metric test and adding the PR's arithmetic/u4-presence record, rows, `pub mod presence` and u4_presence_bits_metrics_are_current fixture. Catalog: main 117 records / 283 configurations, PR 56 / 162, merged 118 / 284 with no duplicates; all main records unchanged. Metric fixture unchanged (1526 bytes, 33 witness bytes, 16 data items, 34-item peak, 936 static non-push opcodes). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Integrate origin/main (3d1001f) into the signed-window schedule branch. - knowledge/negative-results/index.md: keep every main entry (NR-001..NR-066, NR-071, historical PR #3, merged Merkle-branch record) and append this PR's two entries under their assigned IDs. The PR's NR-064 and NR-065 collided with main's NR-064 (data-only signature budgets) and NR-065 (CSV operand narrowing); they are renamed NR-067 (signed-window tables) and NR-068 (width-5 fixed-base CSFS). No PR-side links referenced the old IDs. - src/arithmetic/u4/mod.rs: take main's sorted module list, a strict superset of the PR's alphabetical reordering (47 vs 37 modules, no PR-only module). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…result to NR-069 Integrate origin/main (3d1001f) into the BLAKE3 keyed-mode boundary PR. - knowledge/negative-results/index.md: keep main's file byte-identical (NR-001..NR-066, NR-071) and append the keyed-mode entry under its assigned ID NR-069 instead of the colliding NR-064. Entry text is otherwise unchanged; no PR-side link referenced the old anchor. - src/arithmetic/u4/mod.rs: take main's sorted module list; the PR side only reordered declarations and every PR-side module is present on main. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Run the raw-witness alias regressions as complete leaves under the local TapscriptProfile::Consensus profile that check all 16 presence bits and end with a single OP_TRUE, matching the merged u4 equality and adjacent-equality contract. - compares_nonminimal_numeric_encodings_by_value: [01 00], [80] and [00 00] at every position of a three-item batch, clean-stack checked. - numeric_aliases_follow_ordered_presence_bits: asymmetric witness [01 00], [80], [02 00], 5, [0f 00 00 00] (values 1, 0, 2, 5, 15) with a non-palindromic expected vector; canonical control under Consensus and Policy; the Policy profile rejects the aliases with MinimalData. - byte_equality_and_reversed_output_mutants_fail: test-only mutants (all 80 OP_NUMEQUAL -> OP_EQUAL; reversed output order) fail the same leaf with NumEqualVerify. Add shared scriptnum/run_tapscript/assert_clean_true helpers to arithmetic::test_helpers. Record the new tests, the alias contract and zero hint items in the catalog, knowledge page and u4 README. The production fragment and its metrics are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…icts Integrates origin/main (3d1001f) into feat/sha1-midstate-boundary (94c819e) to address Robin Linus's 2026-09-25 review. Robin's conflict list named only tests/primitive_metrics.rs; the actual trial merge also conflicted on src/arithmetic/u4/mod.rs and src/hashes/sha1/mod.rs. - src/arithmetic/u4/mod.rs: both sides' `pub mod` lists were already a subset of main's, so main's file is taken as-is (verified: every PR mod line is present in main's list). - src/hashes/sha1/mod.rs: kept the PR's new sha1_80bytes_from_midstate continuation function, but adapted it to main's shared u8_reverse_toaltstack() helper (arithmetic::u32::stack) instead of restoring the PR's now-superseded local push_reverse_bytes_to_alt, per the repo convention of keeping main's refactor. - tests/primitive_metrics.rs: kept every #[test] fn and metric key from both sides (134 tests = 133 from main + 1 PR-only sha1_midstate_metrics_are_current; 934 metric keys = union of main's 931 and the PR's 802, no duplicates). All of main's hash metrics (including the new SHA-256 midstate/tagged-hash metrics) remain. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Bring u32_xor_constant tests to the embedded-constant OR/AND/XNOR contract
and make table cleanup explicit:
- rejects_malformed_and_nonminimal_limbs: complete leaf under
TapscriptProfile::Consensus; at all four positions -1/-127/256 -> Verify,
negative zero / non-minimal one -> EqualVerify, 5-byte ->
ScriptIntNumericOverflow; 0..3 limbs -> InvalidStackOperation; canonical
128/255 at every position succeed with the correct XOR; a test-only
top-limb-only mutant accepts the alias at positions 0-2.
- projects_boundary_and_pattern_words: 8 boundary cases plus 100 seeded
cases via arithmetic::test_helpers::{run_with_witness, word_witness}.
- drops_whole_table_and_restores_stack_state (new): OP_DEPTH check after
each of one and three sequential calls with main/alt sentinels; peak 276
for both; a test-only mutant leaving one table item fails the depth check.
- Keep preserves_surrounding_main_and_alt_stack_items.
Helpers stay local, as in and_constant.rs, because sibling helpers are
private. Update the catalog test list and knowledge page; the table is
transient per call, not persistent or reusable. Fragment bytes unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merges origin/main (3d1001f, which has since integrated #84 signed less-than, #45 fixed rotations/equality, #40 reverse-byte adapter, and 27 other PRs) into feat/u32-compressed-boundary, addressing Robin Linus's 2026-09-25 review. Conflicts resolved: - knowledge/catalog.json: kept the u32 record's "tests" entries from both sides (u32_compression_metrics_are_current and u32_reverse_byte_adapter_metrics_are_current), and merged the "security"/"stack_contract" prose so both sides' sentences survive. main ⊆ merged and PR ⊆ merged verified by record-id and (record, configuration)-id set comparison: 117 records / 285 configurations after merge, no duplicates, no ids outside the union of both sides. - knowledge/primitives/u32.md: concatenated both sides' bullets (signed compression boundary / unchecked decoder from the PR; reusable routing / byte-plane transpose / fixed-byte rotations / SHA-256 rotation from main). All 43 main bullets and 40 PR bullets verified present in the merged 45. - src/arithmetic/u4/mod.rs: union of both sides' `pub mod` lines, alphabetically sorted (main added several u4 modules the PR predates; no semantic content lost). - tests/primitive_metrics.rs: kept every #[test] fn from both sides (u32_compression_metrics_are_current from the PR; u32_reverse_byte_adapter_metrics_are_current, u32_equality_metrics_are_current, u32_fixed_rotation_metrics_are_current, u32_byte_planes_metrics_are_current from main). 207 fns after merge = union of main's 206 and the PR's 169 (168 shared). src/arithmetic/u32/README.md and src/arithmetic/u32/stack.rs merged without conflict. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…current main Resolve conflicts against origin/main 3d1001f in knowledge/comparisons/arithmetic.md, knowledge/primitives/index.md, src/arithmetic/u32/README.md, src/arithmetic/u32/stack.rs, src/hashes/sha256/sha2_u32.rs and tests/primitive_metrics.rs by keeping every main row, test and metric plus the PR's u32_not additions. knowledge/catalog.json merged cleanly (117 main records kept unchanged, one PR record added). SHA256: keep main's sha2_u32.rs verbatim (table sharing, input routing and its local unchecked u32_not). The PR's only SHA256 change was a bytecode-identical refactor replacing that local helper with a crate-private u32_not_unchecked; it is not needed, so it is dropped together with the now-unused u32_not_unchecked/u32_not_step helpers. Docs now describe the unchanged SHA256 helper accurately. Add checked_complement_rejects_malformed_and_nonminimal_limbs following or_constant's shared contract (complete leaf, Consensus profile, exact error per malformed encoding at every limb, all-position nonminimal predicate, top-limb-only mutant, canonical 0/127/128/255 controls), and give the pushed-limb loop in checked_complement_rejects_non_byte_limbs a terminal predicate and valid control so it is no longer vacuous. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ical API Resolve conflicts against origin/main 3d1001f in src/arithmetic/u4/parity.rs, src/arithmetic/u4/README.md, src/arithmetic/u4/mod.rs and knowledge/comparisons/arithmetic.md. - parity.rs: keep #233's numeric-range API, docs and all five tests (projects_all_nibble_parities_in_order, distinguishes_asymmetric_output_order, rejects_out_of_range_nibbles, rejects_zero_and_overlarge_batches, respects_combined_stack_frontier) plus the PR's four canonical tests and the separate U4_PARITY_CANONICAL_MAX_BATCH = 981 bound. - mod.rs: main's module list is a superset of the PR's sorted list. - README/comparison: keep rows from both sides; describe the numeric-range and canonical APIs separately. - Knowledge page and catalog: split the numeric-range and canonical-encoding descriptions; the range-only API is not claimed to enforce canonical encoding. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The canonical malformed-input test ran under the default research helper, whose interpreter enforces MINIMALDATA on numeric operands. It therefore kept passing when the canonical byte-equality check was removed from u4_nibbles_to_parity_canonical: the interpreter, not the fragment, rejected the aliases. - canonical_projection_rejects_malformed_nibbles: keep the original loop and add a local TapscriptProfile::Consensus loop (no MINIMALDATA) with a valid control, aliases 0x0100/0x00/0x0f00/0x80 and range failures 0x0001/0x81 at every position, asserting the rejection is not MinimalData. - range_api_accepts_numeric_alias_that_canonical_api_rejects: under the consensus profile the numeric-range API accepts 0x0100 while the canonical API fails with EqualVerify; the default helper rejects it via MinimalData. - canonical_respects_combined_stack_frontier: 979 canonical nibbles with one main- and one alt-stack item reach exactly 1,000 items; 980 fails StackSize. - Qualify docs: numeric-range alias acceptance holds without MINIMALDATA. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…#236 and Core fixture #225 Resolve conflicts in knowledge/catalog.json, knowledge/primitives/u4-lsb.md, src/arithmetic/u4/README.md, src/arithmetic/u4/lsb.rs and src/arithmetic/u4/mod.rs (the last one is new relative to e9d5a66 because main now also carries #244). - lsb.rs: keep #236's range-checked wording, imports and its three tests (projects_all_nibble_least_significant_bits_in_order, rejects_out_of_range_nibbles_at_each_position, respects_stack_frontier_and_preserves_state); keep this PR's separate u4_nibbles_to_lsb_canonical API, its 981-input bound and its four tests. Document the canonical API's contract distinctly. - mod.rs: take main's sorted module list (superset of the PR's). - catalog: keep main's range-only summary/security wording and #225's Core complete-leaf configuration; add the PR's canonical-checked-batch32 configuration, list the canonical tests and metric, and state that only the canonical API enforces canonical ScriptNum encoding. - knowledge page / README: keep #236 and #225 text verbatim; add the canonical adapter as a separate API with its 1..=981 frontier and the note that the Core fixture does not validate it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ality check canonical_projection_rejects_malformed_nibbles ran under the default local helper, which enforces minimal ScriptNum decoding (require_minimal), so the aliases [0x01, 0x00] and [0x80] were rejected with MinimalData by OP_WITHIN before verify_canonical_nibble ran. With the canonicality check removed from u4_nibbles_to_lsb_canonical the test still passed. Run the rejection cases under the local TapscriptProfile::Consensus profile, which decodes non-minimal numbers: add a valid control, a range-only control showing u4_nibbles_to_lsb accepts both aliases there, and exact-error assertions (EqualVerify for aliases, Verify for 256) for the canonical API at every position. The original minimal-helper rejection check is retained. With the canonicality check dropped, the test now fails with "canonical API mishandled [1, 0] at 0" (left: None, right: Some(EqualVerify)). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Integrates origin/main (3d1001f, PR #244's 27-PR batch) into the ripemd160 midstate-continuation branch. Robin Linus's review on 2026-09-25 named tests/primitive_metrics.rs as the integration conflict; the actual merge also produced conflicts in src/hashes/ripemd160/mod.rs and src/arithmetic/u4/mod.rs. - tests/primitive_metrics.rs: kept every #[test] fn and metric `key` from both sides (main's 134 test fns plus the PR's ripemd160_midstate_metrics_are_current, merged set has 135, no duplicates; sha256_midstate_metrics_are_current and sha256_tagged_hash_metrics_are_current preserved as required). - src/hashes/ripemd160/mod.rs: main independently extracted the byte-reversal helper into the shared arithmetic::u32::stack::u8_reverse_toaltstack; adapted ripemd160_80bytes_from_midstate to call that shared helper instead of restoring the PR's now-redundant private push_reverse_bytes_to_alt. - src/arithmetic/u4/mod.rs: main's mod list is a strict superset of the PR's (PR only reordered existing entries); took main's version (47 pub mod lines) verbatim. - knowledge/catalog.json: verified main's 117 records and the PR's 101 records (including the ripemd160-u32 record's message-80-midstate configuration) are both subsets of the merged 117-record file, with no duplicate ids. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The width tests only rejected 2^width. Since 2^width mod 3 is 1 or 2, its final-step accumulator always equals the quotient, so only the remainder branch was exercised: deleting the `acc <= q` check left all tests passing and accepted value 6 at width 2. Add rejects_out_of_range_values_on_both_width_check_branches_at_every_width. For every width 1..=31 it runs a valid 2^width-1 control and requires ExecError::Verify for 58 accumulator-above-quotient values ((q+1)*3 and 3^t-1; the branch is unreachable at widths 1 and 3) and 46 final-trit-above-remainder values. Update the catalog test list, README, knowledge page and research note. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…licts Robin Linus requested integrating current main while retaining #239's checked/unchecked pair-packing and range-error tests on stack.rs alongside this PR's word-transfer and lookup-lifecycle documentation and tests. - knowledge/catalog.json: combined the two independent stack_contract sentence additions (main's popcount wording, this PR's transfer/ lookup wording); kept every record and configuration from both sides (117 records, 292 configurations; main's 283 configs + this PR's 9 new ones). - src/arithmetic/u4/mod.rs: re-sorted the module list, keeping every `pub mod` from both sides. - src/arithmetic/u4/stack.rs: kept both `use` imports (this PR's execute_script_with_inputs_strict and main's bitcoin_scriptexec:: ExecError, the latter needed by #239's tests). All 24 tests in the module (main's 13 plus this PR's 11 new transfer tests) survive; the PR's now-superseded `packs_all_checked_nibble_pairs` is correctly dropped in favor of main's renamed/expanded `packs_all_nibble_pairs`, per instruction to keep main's pair-packing functions unchanged. - tests/primitive_metrics.rs: both sides had appended new tests at the same insertion point; spliced this PR's 4 new u4 metric tests ahead of main's 4 new u32 metric tests, preserving all fns from both sides (206 main + 4 new = 210 total, verified by name-set comparison). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
# Conflicts: # knowledge/catalog.json # knowledge/comparisons/arithmetic.md # knowledge/primitives/index.md # src/arithmetic/u4/README.md
# Conflicts: # knowledge/catalog.json # knowledge/comparisons/arithmetic.md # knowledge/primitives/index.md # src/arithmetic/u32/README.md
# Conflicts: # knowledge/comparisons/arithmetic.md # src/arithmetic/u4/README.md
# Conflicts: # knowledge/negative-results/index.md # knowledge/open-problems.md
# Conflicts: # tests/primitive_metrics.rs
# Conflicts: # knowledge/catalog.json # src/arithmetic/u4/README.md
# Conflicts: # tests/primitive_metrics.rs
# Conflicts: # knowledge/negative-results/index.md
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Integrates 15 reviewed PR heads after their previous review findings were addressed: #245, #136, #135, #132, #130, #116, #113, #92, #75, #73, #72, #71, #49, #42, and #10.
The changes add depth-one Taproot commitment fixtures, checked arithmetic adapters, SHA-1/RIPEMD-160 continuation helpers, focused stack/encoding regressions, and measured research boundaries. Numeric-alias tests distinguish consensus from policy; ternary integer checks reject oversized values before reconstruction overflow.
Resolved overlapping documentation/catalog additions and retained both hash-midstate metric helpers and every existing metric function. Removed three scratch exports introduced by #42 (
knowledge/primitives/u32.md.bullets.txt,tests/primitive_metrics.rs.fns.txt,tests/primitive_metrics.rs.keys.txt). Original PR heads remain ancestors of this branch.Validation:
#62, #26 and #4 remain unchanged and require the integration work documented in their existing review comments.