Skip to content

BDSP AutoStory: the sim's detector misses use the shared fault counters - #429

Draft
dayfine wants to merge 10 commits into
bdsp/sim-faultsfrom
bdsp/sim-misses-faults
Draft

dayfine wants to merge 10 commits into
bdsp/sim-faultsfrom
bdsp/sim-misses-faults

Conversation

@dayfine

@dayfine dayfine commented Nov 12, 2026

Copy link
Copy Markdown
Owner

Depends on #423. Base bdsp/sim-faults; the diff shows only this change.

BDSP's SimMisses picked the nth fade, hold, banner, dark door run and through tap step with its own size_t counters (m_misses.fade == m_misses.fades++). It now counts them on the shared Pokemon::FaultOccurrences from #422, one per miss, as FRLG's sim does in #424. The misses are detection misses with no shared FaultClass (fade or banner unseen, a false dark read, a tap step begun past the warp), so they stay BDSP's own knobs: tests still set them counting from 0, arm_misses() turns them into the counters' chance numbers (from 1), and looked_for() reads how many chances a run saw.

Tested with the headless runner, before and after: PokemonBDSP::AutoStory (story sims, detector-miss Monte Carlo, fault sweeps, sim goldens) gives the same verdicts, 297 passed and the same 5 failing on the base (resources, plan goldens, hand-written legs, overshoot sweep); the golden files are unchanged.

🤖 Generated with Claude Code

https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss

dayfine and others added 9 commits November 12, 2026 05:52
The nth fade, hold, banner, dark door run and through tap step now come
from Pokemon::FaultOccurrences instead of hand-kept size_t counters; the
knobs keep counting from 0, as the tests set them, and arm_misses() turns
them into the counters' chance numbers, from 1. The run's counts are read
with looked_for().

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
…mine plan replay step)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
dayfine added a commit that referenced this pull request Nov 13, 2026
…engine)

Reapplies a7ee7c1 (reverted in dcbd5a4). The shared SimGolden/SimFaults
files are autostory/engine's. FaultSweep_Tests keeps BDSP's own FaultClass:
its START_OFFSET has no shared counterpart yet.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
dayfine added a commit that referenced this pull request Nov 13, 2026
…eads the shared chance counters

route_plans.golden.txt regenerated with PA_UPDATE_ROUTE_PLANS=1 on the lab's
01.02 and 00.05c plans; it now equals lab/dev-bdsp's golden. #429 replaced
SimMisses' fade and banner counts with FaultOccurrences, so the Monte Carlo
test reads their chances().

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant