fix: use closest ancestor parallel slot instead of farthest

[NathanDrake2406]

Summary

• discoverInheritedParallelSlots iterated root-to-leaf but used if (!slotMap.has(slot.name)) which made the farthest ancestor's slot win instead of the closest
• For example, with @sidebar at both app/ and app/dashboard/, the route /dashboard/settings would incorrectly get the root-level sidebar
• Removed the has() guard so slotMap.set() is called unconditionally for inherited slots — since iteration goes root-to-leaf, later (closer) ancestors naturally overwrite earlier (farther) ones
• One-line fix, consistent with the isOwnDir branch which already used unconditional set()

Test plan

• New test with fixture: same @sidebar at root and dashboard levels, verifies /dashboard/settings gets the dashboard-level slot
• All 128 routing tests pass (91 routing + 37 route-sorting)
• Typecheck and lint clean

[pkg-pr-new]

Open in StackBlitz

npm i https://pkg.pr.new/vinext@545

commit: 22d3c03

[james-elicx]

/bigbonk review

[ask-bonk]

Clean, correct fix with a good test. The bug analysis is spot-on: the old if (!slotMap.has(slot.name)) guard caused the first (root-level) slot to win since iteration goes root-to-leaf, when the desired behavior is for the closest ancestor to win. Removing the guard lets later iterations naturally overwrite earlier ones, which is consistent with how the isOwnDir branch already works.

The test creates exactly the right scenario (same @sidebar at root and dashboard levels) and asserts the correct outcome for a leaf route under dashboard.

[ask-bonk]

Nit: You could also assert that sidebarSlot!.layoutIndex matches the dashboard layout (index 1) rather than the root layout (index 0). That would make the test even more precise about which slot was selected, since the fix also ensures layoutIndex is correct. Not blocking — the defaultPath assertion is sufficient to prove the right slot won.

[ask-bonk]

Approved the PR. Here's a summary of the review:

Verdict: Approved

The fix is correct and minimal. The old if (!slotMap.has(slot.name)) guard was inverted in effect — since discoverInheritedParallelSlots iterates root-to-leaf, the guard caused the farthest ancestor's slot to stick (first write wins), when the intended behavior is for the closest ancestor to win. Removing the guard lets later (closer) iterations overwrite earlier ones via slotMap.set(), which is also consistent with the isOwnDir branch that already used unconditional set().

The test is well-constructed: it sets up the exact scenario (same @sidebar at both root and dashboard/ levels) and verifies that /dashboard/settings gets the dashboard-level slot. I left one minor suggestion about additionally asserting layoutIndex for extra precision, but it's non-blocking.

github run