[Musab1258]

Description

compile --type tr put the whole policy into a single taproot leaf, so spending any branch of an or revealed every other branch on-chain. This gives each spending path of a disjunctive policy (or, thresh(1,..)) its own leaf, arranged as a probability-weighted Huffman tree, so only the path that is used is revealed.

The internal key is unchanged: it is still the randomized unspendable key H + rG, and r is still returned (#218, #225). Conjunctions, thresh(k,n) with k > 1, and single keys still compile to one leaf.

or(pk(A),pk(B)) before and after:

before: tr(<H+rG>,or_b(pk(A),s:pk(B)))
after:  tr(<H+rG>,{pk(B),pk(A)})

Closes #244.

Notes to the reviewers

Why not compile_tr? Its extract_key() picks a policy key as the internal key whenever one can satisfy the policy alone, and the unspendable key is only a fallback. For or(pk(A),pk(B)) it returns tr(A,pk(B)), so r would no longer prove anything about the internal key. It also returns Empty Miniscript compilation for pk(A) on miniscript 12.3.7 and 12.3.8 (rust-miniscript#902). So the tree is built here from public miniscript APIs, and only the internal key comes from the existing code.

How it works. compile_tap_tree in src/utils/descriptors.rs:

1. Validates the whole policy first (is_valid() plus the safety checks, as compile_tr does).
2. Flattens root-level or and thresh(1,..) into (probability, branch) pairs.
3. Compiles each branch with compile::<Tap>() and runs sanity_check().
4. Combines the leaves into a Huffman tree with TapTree::combine, capped at 1024 leaves (miniscript's limit). Ties break deterministically.

The only change to compile_policy is the line that built the single leaf. It works for both DescriptorPublicKey and String (--placeholders) keys.

Whole-policy validation matters. Validating each branch alone accepted or(pk(A),and(pk(A),pk(B))), which wsh rejects with "duplicate keys". Step 1 prevents that, and there is a test for it.

Trade-off. Unlike stock compile_tr, or(A,B) gets no key-path spend through A. That follows from keeping the verifiable unspendable key required by #218.

Commits. There are two:

1. Tests only, covering behavior that must not change. They pass on the current code.
2. The change, its tests and the CHANGELOG entry.

Testing.

• Commit 1 adds 6 integration tests: single key, conjunction and thresh(2) stay one leaf, a key reused across branches is rejected, an unsafe branch is rejected, and the internal key is the unspendable key with r returned.
• Commit 2 adds 4 integration tests (split or, split thresh(1), 1024 leaves accepted, 1025 rejected) and 8 unit tests on tree shape, including that a likely branch sits nearer the root.
• The existing test_compile_* tests are untouched and pass.
• Run against the old code, the 4 new-behavior integration tests fail and the guard tests pass.

Not covered.

• The taproot depth limit (128) with a heavily skewed tree is untested. Descriptor::new_tr should reject it.
• Expanding thresh(k,n) into and leaves (the compile_tr_private_experimental behavior) is left for a follow-up.
• PR refactor(compile): extract unspendable key derivation #368 adds verification of r. Since this PR does not touch the internal key, it should compose with refactor(compile): extract unspendable key derivation #368. I'll add a test that runs verify_unspendable_internal_key on split trees once refactor(compile): extract unspendable key derivation #368 merges.

Checklists

All Submissions:

• I've signed all my commits
• I followed the contribution guidelines
• I ran cargo fmt and cargo clippy before committing

New Features:

• I've added tests for the new feature
• I've added docs for the new feature

[codecov]

Codecov Report

❌ Patch coverage is 98.92473% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 60.74%. Comparing base (3dec0e0) to head (4b64f52).
⚠️ Report is 3 commits behind head on master.

Files with missing lines	Patch %	Lines
src/utils/descriptors.rs	98.92%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #378      +/-   ##
==========================================
+ Coverage   60.00%   60.74%   +0.74%     
==========================================
  Files          25       25              
  Lines        4736     4828      +92     
==========================================
+ Hits         2842     2933      +91     
- Misses       1894     1895       +1     

Flag	Coverage Δ
rust	60.74% <98.92%> (+0.74%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Musab1258]

@vadim-anfv, can you please check this out when you are free?