Same-value branch switch leaves a synchronous memo inconsistent with its inputs

[GabbeV]

Describe the bug

While B is held at 0, updating A to 2 leaves Selected showing 1.

Selected is b() ? b() : a(), so with the displayed inputs A=2 and B=0, it should select A.

Your Example Website or App

https://s.olid.uk/id/qbPG6OYNRoG2FZgYcpia-Q

Steps to Reproduce the Bug or Issue

1. Wait for initialization.
2. Click Run.
3. After half a second, A shows 2, B shows 0, and Selected still shows 1.

Expected behavior

Selected should remain consistent with the published inputs. Either A should remain held, or Selected should update alongside it.

Screenshots or Videos

No response

Platform

• OS: Windows
• Browser: Firefox

Additional context

Related to #3461, fixed by #3464. This version uses a synchronous memo, and switching branches initially produces the same value (1). A subsequent update to the previously selected source exposes the inconsistency.

Reproduces on next at 347a5ca.

[ryansolid]

Fixed in 5fa224a via #3479 (GabbeV hold-consistency batch 2).

Rule: A30. heldTrims: an unchanged pass's dependency trim waits on the flush's verdict (drained at commit, dropped on park); creation, OPT-dirty and tracked-effect passes trim at once. The memo arm resolves to the held outcome (A: 2 reveals with B: 1), the pre-existing sum = a() + b() behavior; the render-effect arm follows its inputs at once.

— Claude via Cursor