Repository navigation
Conversation
A frame source serves a recording's frames ("<ms>.png") by virtual time,
and replay_until / replay_events feed them to VisualInferenceCallbacks
or a ScreenEventObserver at the inference period on one thread, so a
console failure can become a deterministic test of the real detectors.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
…ames The walk's warp watcher and FRLG's screen events, replayed over each run's end frame (CommandLineTests/PokemonFRLG/FrameReplay/), agree that no fade showed: the walk ended off the door, not past a missed fade. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
…stration function at 50 lines Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
…ents' tests PokemonFRLG_Tests.cpp is at its 1000-line limit. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
The late-watcher case showed a fade missed by a watch that starts after the hold; it now replays the walk's watch (WatchAlong over the hold, then the same watcher on) and checks it sees a door's black that is over before a 2000 ms hold ends, while a watcher started at the end of the hold still misses it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
…atched along the hold)
replay_watcher() shows the hold's frames to the walk's BlackScreenOverWatcher through a WatchAlong, then waits on with it, as walk_overworld_route() does since its fade watch covers the hold; a fade over before the hold ends reads "fade over during the hold". Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Depends on #293 (base
frlg/stage8-rocktunnel) and #306 (merged in); the diff shows #306's files too until it lands in the base.Replays the 00.27
route10_nursefailures ("no fade in and out after walking onto the warp to Route10_PokemonCenter_1F", console B 2026-11-09, runs 2 and 3) through the walk's real warp watcher and FRLG's screen events.BlackScreenOverWatcherafter a one-tile hold, 8 s) andFrlgScreenObserver+judge_warp()/classify_interrupt()both say: no fade, no interrupt. The real detectors were right. The walk ended off the door (Route10 (14, 10) and (16, 19)), which matches the map-evidence notes (timed legs one tile off), so the fix belongs in the route, not the detectors.CeruleanCityone. The frames live outside the repo at~/FRLGLab/captures/frame-replay/CommandLineTests/PokemonFRLG/FrameReplay/0027-route10-run{2,3}/0.png(1080p, 2.2-2.5 MB each); the cases skip without them.PokemonFRLG_Tests.cppis at its 1000-line limit.Testing
Headless
SerialProgramsTests(scratch merge withupstream/headless-tests), filtersPokemon::,PokemonFRLG::AutoStory - Screen,Frame replay, with the fixtures linked asbuild/CommandLineTests: 208 passed, 0 failed, 2 skipped; both 00.27 replay cases ran (not skipped). Code health: 0 failures.🤖 Generated with Claude Code
https://claude.ai/code/session_013VUubhHeYSMGSSDXZkKZss