Glitch E: seeded starts and Peep Show

Both seeded start failures are fixed. The error panel's desktop width was already fixed on current main. Conjoined Peep Show now sends the private choice and card to the shared hand owner.

Base 42b063ee · Candidate 28b3c1fd8f9268bb7cf371be5352dbb00f9a1514 · Draft PR #229

The old dealer searched 512 seeds and could find no deal for either reported pair. The new setup arranges real copies in the opening deck, then runs the normal round-robin deal. It retains the supplied seed, deck counts, deal events and ordinary rules.

V36: Market Crash + Corporate Bankruptcy

Seed dm6, deal market-crash,corporate-bankruptcy, seat 0. Before: Could not start. After: round 1 opens with both requested cards.

390 × 844

before V36: Market Crash + Corporate Bankruptcy at 390 by 844 — Before

after V36: Market Crash + Corporate Bankruptcy at 390 by 844 — After

1440 × 900

before V36: Market Crash + Corporate Bankruptcy at 1440 by 900 — Before

after V36: Market Crash + Corporate Bankruptcy at 1440 by 900 — After

1920 × 1080

before V36: Market Crash + Corporate Bankruptcy at 1920 by 1080 — Before

after V36: Market Crash + Corporate Bankruptcy at 1920 by 1080 — After

Related start failure: Gridlock + Blind Box

Seed gl3-blind-box, deal gridlock,blind-box, seat 0. Before: Could not start. After: round 1 opens with both requested cards.

390 × 844

before Related start failure: Gridlock + Blind Box at 390 by 844 — Before

after Related start failure: Gridlock + Blind Box at 390 by 844 — After

1440 × 900

before Related start failure: Gridlock + Blind Box at 1440 by 900 — Before

after Related start failure: Gridlock + Blind Box at 1440 by 900 — After

1920 × 1080

before Related start failure: Gridlock + Blind Box at 1920 by 1080 — Before

after Related start failure: Gridlock + Blind Box at 1920 by 1080 — After

Error panel width: already fixed

The baseline captures above show a 460px column at 1440 × 900 and 1920 × 1080. At 390 × 844 it is 358px wide with 16px side margins. No UI files changed in this PR.

Conjoined Peep Show: engine replay

Player a is the caster and bankroll member. Player c owns the shared hand. Both runs offer Diamond Armor, Void and Invisible. Picking Void gives these actual engine results.

Version	Chooser	Card recipient	a's set-aside hand	c's shared hand
Before	a	a	2	1
After	c	c	1	2

The corrected hand-owner bot still chooses Void through its existing card-ranking rule. Regression tests cover player answers, bot answers, timeouts and the two unused cards going to the bottom of the deck. The resolver replay isolates the non-hand caster case; it is not a browser capture.

Replay data

[
  {
    "phase": "before",
    "caster": "a",
    "handOwner": "c",
    "chooser": "a",
    "offered": [
      "diamond-armor",
      "void",
      "invisible"
    ],
    "privateTo": [
      "a"
    ],
    "botAction": {
      "type": "answer",
      "promptId": "choice-1",
      "value": {
        "kind": "cards",
        "uids": [
          "void-4"
        ]
      }
    },
    "picked": "void",
    "pickRecipient": "a",
    "setAsideCount": 2,
    "sharedCount": 1
  },
  {
    "phase": "after",
    "caster": "a",
    "handOwner": "c",
    "chooser": "c",
    "offered": [
      "diamond-armor",
      "void",
      "invisible"
    ],
    "privateTo": [
      "c"
    ],
    "botAction": {
      "type": "answer",
      "promptId": "choice-1",
      "value": {
        "kind": "cards",
        "uids": [
          "void-4"
        ]
      }
    },
    "picked": "void",
    "pickRecipient": "c",
    "setAsideCount": 1,
    "sharedCount": 2
  }
]

Verification

Typecheck, mobile lint/i18n guard, 6 focused engine checks, all 12 FX sample/deal checks and all 51 bot checks passed. New regression cases were observed failing before the fix. Final gate result is recorded in the PR Notes.

Baseline browser data · Fixed browser data · Engine replay data

Engine change: the coordinator must redeploy the game server after merge. Independent review and stamping belong to the coordinator.