Repository navigation
Conversation
03.08's catch_more_species walked out of Wild Zone 1 (console B, 2026-11-16): the walk to the 100 stop read 109, then the marker's digits did not read for 13 s while it kept stepping, and it stopped on the first reads it took again, 37, in the city by a market stall. No Pokemon locked on there, so ten stops threw nothing. The badge 20 s in was the game's autosave on leaving the zone. MarkerDistance (replacing DistanceReads) counts each 0.6 s step since the last read taken as 3 of the distance, so a walk to a stop ends there when the digits stop reading. The 03.08 plan notes the autosave; it keeps the catches' autosave as its save.
03.08 run 2 (console B, 2026-11-16) read a lone "1" at about 135 on the walk's first look, took it as the start and ended the walk to 100 at once. A read is now taken when it is near the position (the last read moved on by the steps since), or when two of the last three reads not taken agree with it, each moved on by its steps; so the first read needs three that agree, and a lone misread among them does not hold the start back.
… MarkerDistance 03.08 run 3 read 1, 1 and 4 with the marker at 120 behind a rock; those agreed within 12 and ended the walk to 100 at once. A read not taken that, moved on, is at 0 or less now agrees with nothing. The catch steps' start read (Taunie's rocks, the walk back to the road) goes through MarkerDistance too, and the lines after a throw press B, not A, on an A prompt over the scene (a zone gate's "A Leave").
03.08's walk to 100 stepped on through 13 s of unread and garbage digits after 109 and came out at 37 in the city (console B, 2026-11-16). Dead reckoning is gone: the walk steps only on a distance read taken, and with none it stands, turns the camera back onto the marker and reads again. The patrol's walks toward the objective never go below 57 (60, just before the far gate, less the stop slack), and reads below it are not taken, so the lone 1 and 4 at Taunie's rocks (runs 2 and 3) give no start. The A-prompt B guard in press_a_through_lines is dropped: the walk out had no A press. The 03.08 plan drops the zone-exit autosave note: that badge was a false one on the zone walls; its start save is 03.07's after_catches autosave.
…locked Run 4 of 03.08 (console B, 2026-11-16) stood with the marker's digits under an NPC's speech bubble; re-centering put the marker right back under it. MarkerDistance now answers every third look in a row with no read taken with a new view: the camera nudged left, then right (the marker kept 0.12 of the width off the middle, by its own x), then a short sidestep across the line to the marker, left then right, out to two steps each side. After those it is BLOCKED and the walk ends, logged, without a step on. Every nudge and sidestep is logged.
…lking back Run 5 of 03.08 (console B, 2026-11-16) ended 7 sidesteps left, against a wall off Wild Zone 1's road: sidesteps were never undone and each stretch started left of where the last left off. MarkerDistance now gives the lane across the marker's line (1 left, 1 right, 2 left, 2 right, around the line), back to 0 after each read taken and when the walk ends; the walk steps to that lane. The walk back from 60 read 65 on every look for 15 steps back, twice, and ran out of time: the reads kept coming, the walk was held up, and held-up detection skipped walks back. MarkerDistance now treats 4 reads taken with none further on, in either direction, as held up and sidesteps to the next lane, kept until the walk is 6 further on.
catch_in_zone no longer walks between marker-distance stops: it turns the camera in 8 steps with a ZL lock-on after each, throws at species not caught, and takes one short leg only after a full turn finds nothing new (each leg out walked back; none toward the objective at an agreed read of 60 or less; a pause for spawns after each round). ZoneSearch holds the state and is tested by replayed scans.
Run 9 searched at the rocks, where the objective marker never read: two full turns with no name plate. 03.08 now starts with ZoneStart: steps west by the minimap (face_heading(), moved into the story backend so catch_in_zone() can use it) until a marker read agrees, then the marker walk to the road's spot at 100. It stops as "rocks_not_left" or "zone_spot_not_reached" before any search. The search's legs go out from 100, and a marker walk back to it follows each catch, fight and round.
Owner
Author
|
Superseded by #644 (one lean branch; split for review later) |
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 #619 (base plza/0304-save-moved; the diff shows only this change).
03.08's catch_in_zone walked Wild Zone 1's road between two marker-distance stops (100 and 60). Six runs on console B (2026-11-16) failed that way: the digits read nothing for long stretches (NPC speech bubbles, the marker at the screen's edge; ~87-101 on the road gave no reads at all), each 30 s walk was eaten by the blocked-read nudges and sidesteps and stuck at ~87-89, and run 6 ended "short_of_species: 4 of 5" with only species already caught locking on at its stops.
catch_in_zone now stays where it is and searches: it turns the camera in 8 steps with a ZL lock-on after each, throws at a target whose plate shows no caught mark, and fights a wild battle that is on. Only a full turn with nothing new takes one short leg (2 s, the camera on the marker); each leg out is walked back before the next way out, so the search stays within a leg of Taunie's rocks, and no leg goes toward the objective when an agreed marker read is 60 or less. After a round of legs it stands 15 s for spawns. The marker's digits only feed that floor; they never steer. Each scan step's result and each leg (way, distance read, offset in legs) is logged. The party bar still decides when it is done; 10 minutes and 12 throws at most, then short_of_species. The new ZoneSearch holds this state machine; walk_back_to_zone_road and the MarkerWalk/MarkerDistance code are unchanged.
Tested with the new ZoneSearch suite (a new species on the third step is caught, a catch or battle restarts the scan, an empty turn walks one leg, legs stay within one leg of the start over two rounds, no leg toward at a read of 55, legs toward at 61) and syntax-checked locally; not run on the console yet. Open on the console: whether 8 steps of 300 ms full right stick make a full turn, and whether a ZL with no target recenters the camera between steps.