Skip to content

battle_bushido fails at seed shift 6 on main: a single A edge against a window the battle may have taken away #229

Description

@mtklein

Found while seed-sweeping the #227-family fix to battle_bushido (55cdf83), and confirmed pre-existing: unmodified main fails identically.

FAIL: assertEq failed: row 0 banked boost 1 ($3e9d = 1): got 0 ($0), want 1 ($1)

at frame 774 — before anything that change touches is reached. Shifts 0/1/2/4/5 pass.

The shape is the same family as the three already found this cycle: pressRowOnce puts a single A edge against a window that the battle may have taken away by then, so whether the press lands depends on timing the test does not control. It is luck standing in for a precondition.

Worth fixing the same way the others were — drive to the state rather than press once at it — but kept separate from 55cdf83 rather than folded in, since it fails without that change too.

Related: #227 (the sfigaro settle), ed0cb36 (the rare-drop precondition), and the silent-green chip arm fixed in 55cdf83.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingharnessTest harness, audits, runner, graph

    Projects

    No projects

      Milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions