bug: next-to-act flash stuck on the last folder when their fold closes a called all-in (fold + runout + hand_ended in one snapshot)

Open
  • Is something wrong?
#968 1 comment 0 reactions 1 assignee View on GitHub

Maintainers usually reply within 1 day

@bitcoinbrisbane is already working on this.

Since Nov 6, 2026.

Assessment

This issue has not been assessed yet.

Description

animation bug game-flow sit-and-go UI Week 2

Symptom

Wednesday 2026-11-04 Texas HODL $2 SNG, table 0xaafb3807d3c36469fe5423d37eae670fd066f938f7e0042beb80ab295065e1db, hands 10 and 12. In both hands a player went all-in and got called, with other player(s) still to act behind. After the last player folded, the UI kept the flashing next-to-act indicator on that player as if the table was still waiting on them — but their fold had committed on chain. Reported by the players at the table; the folder themselves confirmed they folded.

Chain evidence — the chain is correct, this is a rendering bug

Action timelines reconstructed from action_performed events (tx_search on node.texashodl.net; full query at the bottom).

Hand 10 (blocks 898333–898350) — flop bet, all-in call for less, third player folds, fold+runout+hand_ended all in ONE tx:

idx=300 h=898348 b521s8aug…  bet   900        tx=5B10860838
idx=301 h=898349 b5219dj7…   call  15 (all-in for less)  tx=81E86B0128
idx=302 h=898350 b5210fw5…   fold             tx=67512FAA90  ← hand_ended fires in THIS SAME TX

Result: full runout Th 2c Ad Qd Ah, showdown, 1 winner. The fold at idx=302 closed the betting, so the engine dealt turn+river and ended the hand inside one snapshot.

Hand 12 (blocks 898373–898381) — preflop all-in raise, call, two behind; the last fold again lands in the same tx as hand_ended:

idx=331 h=898374 b521p78923…  raise 580       tx=7594DDADE1
idx=337 h=898379 b521p0tkyw…  call  580       tx=6104521BCF
idx=338 h=898379 b521fh8ke9…  fold (SB)       tx=2CFA733305
idx=339 h=898381 b5219dj7…    fold (BB)       tx=388939138B  ← hand_ended fires in THIS SAME TX

Common fingerprint: the final fold both (a) closes the betting against a called all-in and (b) triggers the all-in runout + hand_ended in a single chain snapshot. The turn indicator then stays flashing on the seat that folded.

Suspect

This is exactly the single-frame scenario expandFrames (src/bus/expandFrames.ts) splits into per-street synthetic sub-frames, and the runout slow-down shipped in #873 / #874 (closed 2026-10-30, days before this game) changed the pacing of that path. Likely mechanisms to check:

  1. The synthetic sub-frames (projections between prev and next) carry a stale nextToPlay-equivalent pointing at the folder, and the seat-highlight (utils/turnClock.ts, components/playPage/Table/components/PlayerSeating.tsx + seat/Player components) renders it during the held runout choreography.
  2. The turn indicator reads the rendered track while the hold (showdownHold / runout minDisplayMs) is active, and nothing clears the "waiting on seat N" state when the terminal frame's fold is part of the same commit.
  3. A derive-events gap: deriveEvents(prev, next) may not emit a playerActed/fold transition for the folder when the same diff also contains street advances + winners, so whatever state the flash keys off never updates.

What exists today (file paths)

  • Bus runout expansion: ui/src/bus/expandFrames.ts (+ docs/plans/2026_09_09_runout_frame_expansion.md)
  • Event derivation: ui/src/bus/deriveEvents.ts
  • Turn/flash rendering: ui/src/utils/turnClock.ts, ui/src/utils/timerUtils.ts, ui/src/components/playPage/Table/components/PlayerSeating.tsx
  • Related prior art: #57 (turn ring with no active hand, closed), #873/#874 (runout pacing feature this likely regressed from)

Repro / test plan

  1. Playwright + PVM stub (packages/e2e): inject prev snapshot = 3-way pot, player A all-in, player B all-in-call-for-less, player C to act; inject next snapshot = C folded + full runout + winners (one frame, as the chain sends it). Assert the next-to-act flash is NOT on seat C during or after the runout choreography (window.__B52_BUS__.commitLog for frame assertions).
  2. Unit-test expandFrames for this input: every synthetic sub-frame must carry a cleared/-1 next-to-act once the folder's action is part of the source diff.
  3. Same scenario via deriveEvents: assert a fold event is emitted for C even when the diff spans fold→runout→winner.

Raw data

GAME=0xaafb3807d3c36469fe5423d37eae670fd066f938f7e0042beb80ab295065e1db
# per-hand actions (hands 10, 12):
curl -G "https://node.texashodl.net/rpc/tx_search" \
  --data-urlencode "query=\"action_performed.game_id='$GAME' AND action_performed.hand_number='10'\"" \
  --data-urlencode per_page=100 --data-urlencode "order_by=\"asc\""
# indexer hand records:
curl "https://indexer.block52.xyz/api/v1/hands/$GAME/10"
curl "https://indexer.block52.xyz/api/v1/hands/$GAME/12"
Dominant language
TypeScript
Stars
2
Forks
5
Avg merge
21h 38m
Merged PRs (30d)
105

Getting set up

We have not checked this project's setup files yet. Start from its README, and see our first-contribution guide for the general steps.

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from block52/ui

All issues in block52/ui

Similar issues

More TypeScript issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.