[johnml1135]

Start here: src/SIL.Machine/Annotations/AnnotationList.cs, FindDepthFirst (+41 lines): the whole fix.

Ready for review. All CI checks pass.

AnnotationList.FindDepthFirst now tries every container that shares the range of the one Find returned. It used to descend only into that single container, so a segment inside an equal-range sibling was found or missed depending on insertion order and skip-list levels. The failing-lookup result (nearest descendant) is unchanged.

Reviewers arrive asking whether this changes lookups that already work. It does not: with one container at a range the group has one member and the path is the old one. The question worth your time: when several tied containers each own the offset, the walk returns the first match, trying the Find result, then earlier tied nodes, then later ones.

Where to look:

• FindDepthFirst and the new private GetSameRange: walks the equal-range group in both list directions, so it does not depend on which tied node Find picks. Independent of the BidirList.Find boundary change in Fix directional annotation tie boundaries #556.
• AnnotationTiedContainerTests: four tied containers, the owner at each position, six fixed skip-list seeds (set by reflection on the test instance only), both search directions. Pins flat ties, nested ties, and ties that do not own the offset.

Not here:

• The directional tie boundary in BidirList.Find (HermitCrab: investigate unstable zero-width morpheme identities across fresh processes #506, Fix directional annotation tie boundaries #556).
• Flat spanning-annotation end-removal in Shape.Remove (Shape.Remove can leave a spanning annotation anchored to a removed end segment #555).
• Equal-range handling in GetNodes, Shape.UpdateAnnotations and Matcher: outside AnnotationList.FindDepthFirst misses descendants in tied containers depending on insertion order #554's acceptance criteria.
• No HermitCrab-level reproduction is claimed.

Verification: ./local_check.sh --agent-strict passed (comment hygiene, csharpier, build, tests). With only AnnotationList.cs reverted to master, most of the new test cases fail. Not run: this branch combined with #556.

Next: approve, or say which tied-ownership case is missing from the tests.

---

Reading this a year from now -- start here

Fixes #554 (open): FindDepthFirst missed descendants of tied containers depending on insertion order. Related open items: #506, #555. #500 and #556 are open PRs touching neighbouring tie behaviour. The fix is deliberately limited to AnnotationList.cs; the reasoning below is the only record.

Decisions, and why

• Walk the equal-range group in both list directions instead of changing which tied node Find returns: that choice is what Fix directional annotation tie boundaries #556 changes, and this fix must hold whichever node Find picks.
• Failing lookups still return the nearest descendant, so callers that rely on the miss result see no change.

Evidence

• ./local_check.sh --agent-strict: passed (comment hygiene, csharpier, build, tests).
• AnnotationTiedContainerTests passed with the fix in the earlier run; test case counts were not recomputed for this description.
• With only AnnotationList.cs reverted to master, most of the new cases failed and a minority passed.
• Not run: this branch combined with Fix directional annotation tie boundaries #556.

🤖 Generated with Claude Code

---

This change is

[codecov-commenter]

Codecov Report

❌ Patch coverage is 86.20690% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 74.58%. Comparing base (74ee09b) to head (d636b09).
⚠️ Report is 1 commits behind head on master.

Files with missing lines	Patch %	Lines
src/SIL.Machine/Annotations/AnnotationList.cs	86.20%	3 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #557      +/-   ##
==========================================
+ Coverage   74.57%   74.58%   +0.01%     
==========================================
  Files         453      453              
  Lines       38314    38341      +27     
  Branches     5280     5286       +6     
==========================================
+ Hits        28573    28597      +24     
- Misses       8576     8579       +3     
  Partials     1165     1165              

☔ 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.

[claude]

FYI: F4. Because _rand is set by reflection, renaming that private field would make all 252 cases fail with a NullReferenceException from the !. SIL.Machine already declares InternalsVisibleTo("SIL.Machine.Tests"), so an internal seed constructor on BidirList would give the tests a typed hook instead.

[claude]

Verdict: approve. d636b09 removes the second subtree search on a miss and keeps the old miss result.

Most important: nothing blocks. F4 (tests/SIL.Machine.Tests/Annotations/AnnotationTiedContainerTests.cs:17) is the only open finding. It is Low, so it needs no answer.

Counts: 0 Critical, 0 Important, 1 Low open. Nothing new this round.

Ran: git diff 04b7986 d636b09 and read the new loop. GetSameRange yields candidate first, and Annotation does not override equality, so miss is the old miss result. dotnet test filtered to SIL.Machine.Annotations: 257 passed. CI per gh pr checks 557: comment hygiene passed, while Build on ubuntu-22.04 and windows-latest is still pending.

Not verified: this branch combined with #556.

Public API, target framework, package, machine.py parity: None verified.

F1 withdrawn (the miss path is unchanged from master). F2 addressed. F3 withdrawn (the wider-sibling case is unchanged from master). F4 open.

Reviewed at d636b09