Fix edit algorithms deleting the wrong child via remove_child(<pointer>)

[TroyHernandez]

Composition only declares remove_child(int index, ErrorStatus*), but overwrite() and insert() in editAlgorithm.cpp call it with a Composable* / Retainer:

composition->remove_child(transition);     // overwrite() and insert()
composition->remove_child(items.back());   // overwrite()

The pointer argument decays to bool (always true) → int(1), so these calls remove the child at index 1 (or, via adjusted_vector_index, the last child when size() == 1), not the item that was passed.

Impact

• overwrite() over a multi-item span deletes the wrong clips: the partially-overwritten clip is removed while a fully-overwritten clip survives in place.
• Transition removal in overwrite()/insert() removes index 1 instead of the intended transition.

Repro

Track A|M|B (each 5/8/5 frames), overwrite(X, track, [8, 18)) — a span that partially covers M and fully covers B to the track end:

before (buggy):  A[0+5] B[20+5] X[2+10]   # M removed, B kept — wrong
after  (fixed):  A[0+5] M[10+3] X[2+10]   # M trimmed, B removed — correct

Fix

Pass the already-computed (and bounds-checked) index for transitions, and index_of_child(items.back()) in the removal loop.

Tests

• Adds test_edit_overwrite_partial_first_full_last, which fails before this change and passes after.
• The existing test_editAlgorithm suite continues to pass.

I verified the fix and the regression test by building the library locally (the test fails on the unfixed tree, passes on the fixed tree).

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: TroyHernandez / name: Troy Hernandez (8680cd2)

[codecov-commenter]

Codecov Report

❌ Patch coverage is 20.00000% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 84.33%. Comparing base (9d3c3a1) to head (6670f49).

Files with missing lines	Patch %	Lines
src/opentimelineio/algo/editAlgorithm.cpp	0.00%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2025      +/-   ##
==========================================
- Coverage   84.33%   84.33%   -0.01%     
==========================================
  Files         181      181              
  Lines       13267    13268       +1     
  Branches     1214     1214              
==========================================
  Hits        11189    11189              
- Misses       1895     1896       +1     
  Partials      183      183              

Flag	Coverage Δ
py-unittests	84.33% <20.00%> (-0.01%)	⬇️

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

Files with missing lines	Coverage Δ
src/opentimelineio/serializableObject.h	83.78% <100.00%> (ø)
src/opentimelineio/algo/editAlgorithm.cpp	0.00% <0.00%> (ø)

---

Continue to review full report in Codecov by Harness.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9d3c3a1...6670f49. Read the comment docs.

🚀 New features to boost your workflow:

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

[jminor]

Oh, good catch!

Would you mind adding explicit to the operator bool() on Retainer? I think that would have caused this error to be caught by the compiler & will help prevent folks using OTIO from falling into this same trap.

(cc @meshula to check this)

[TroyHernandez]

Done in 6670f49 (also merged main to pick up #2020, which touches the same functions).

• The library, all C++ tests, and the Python bindings compile with zero call-site changes; full CI matrix is green.
• I verified the guard works: compiling the original bug pattern, composition->remove_child(some_retainer), now fails with invalid user-defined conversion from 'Retainer<Composable>' to 'int'.
• For @meshula re: compatibility, the break surface is narrower than "explicit operator bool" usually implies. Since operator T*() is still implicit, boolean contexts (if (r), &&, !) and plain bool b = r; all still compile via the pointer conversion. The only thing explicit forbids is using bool as a bridge to another type (Retainer -> bool -> int), which is exactly the trap this PR fixes. I'd expect zero impact on correct downstream code.

[meshula]

Thanks!

[meshula]

Thank yo for this!!