Skip to content

PokemonLZA AutoStory: Main Mission 01 checkpoint plans - #565

Draft
dayfine wants to merge 9 commits into
plza/upstream/p3-02from
plza/upstream/m1-01
Draft

dayfine wants to merge 9 commits into
plza/upstream/p3-02from
plza/upstream/m1-01

Conversation

@dayfine

@dayfine dayfine commented Nov 15, 2026 •

Copy link
Copy Markdown
Owner

Depends on #564

Adds the checkpoint plans of Main Mission 01 as data: 01.00 (a new game from Confirm Choices to the station platform), 01.01 (the platform to the mission), 01.02 (Pancham's chase) and 01.03 (the starter and Andi's battle). Each file lists the checkpoint's steps, its start place and its save. Nothing reads them until #568 builds the plan files into the program and its table tests (PokemonLZA::AutoStory_StoryTable) lint every step against the game's actions.

@dayfine

dayfine commented Nov 15, 2026

Copy link
Copy Markdown
Owner Author

API review (claude-sonnet-5-5) at aeddc3b: LGTM_with_nits

Data-only change with no correctness bugs visible from the diff. The why and source strings carry lab-run and history talk (console B, dates, 'was the recorded walk', 'split from ...'), which belongs in the PR description. Some why text is also unclear, and one step has a placeholder duration. I could not check the actions or schema because the table tests land in #568.

  • [style] SerialPrograms/Source/VideoGames/PokemonLZA/Programs/AutoStory/StoryData/plza-01.02.plan.json:14 The why for chase_pancham mixes what the step does with lab and history talk: 'Measured ... on console B', '(console B, 2026-11-13)', 'since 01.01's scenes leave the player turned...'. It also says 'NEEDS TUNING' and that the sidestep's 1000 ms (the [90, 1000] leg) is a placeholder. The placeholder is easy to lose in a long string. The [-90, 0] zero-length leg is unexplained, and the why says the first turn is 'L first' while the first leg is -15. Fix: Keep only what the step does and why it is built this way: walk after Pancham, through the archway, around the starters. Explain the zero-length -90 leg (a turn?) and the leg angles. Move the console, date and 'placeholder' notes to the PR description or a tracked issue. If the placeholder stays, mark it in a structured field rather than inside the prose.
  • [style] SerialPrograms/Source/VideoGames/PokemonLZA/Programs/AutoStory/StoryData/plza-01.01.plan.json:23 The why describes this PR's history ('6000 ms was the recorded walk, which overshot into it', 'Measured 2026-11-08 on console B') instead of what the screen does. The leg values [0,4500] and [90,10000] are never justified. Fix: Say what the step does: walk down the platform for ~4.5 s, then right until the white fade, with 10 s as the upper bound before the fade. Drop the console, date and earlier-value history.
  • [style] SerialPrograms/Source/VideoGames/PokemonLZA/Programs/AutoStory/StoryData/plza-01.03.plan.json:14 The why repeats itself ('only Chikorita's walk was recorded' and 'Only Chikorita's walk is recorded'). It also contains lab-run talk ('Measured ... on console B and checked by hand'). 'walk not run as a story step' is unclear. Fix: Write one sentence: walk up to Chikorita and stop at the 'A Talk Chikorita' prompt. Say once that only Chikorita's path exists. Remove the console and verification-history notes.
  • [nit] SerialPrograms/Source/VideoGames/PokemonLZA/Programs/AutoStory/StoryData/plza-01.03.plan.json:46 The metadata source is provenance trivia, not data: 'console B recording', dates, and a chain of 'split from 01.01 at ... split from 01.02 at ...'. The same pattern is in the other three files' source fields (plza-01.00 line 26, plza-01.01 line 39, plza-01.02 line 44). Fix: Drop the console and split-history from source, or reduce it to a short stable label. Put the history in the PR description or commit message.
  • [nit] SerialPrograms/Source/VideoGames/PokemonLZA/Programs/AutoStory/StoryData/plza-01.03.plan.json:49 This checkpoint saves with 'menu', while the other three use 'autosave'. The last step's why mentions the Main Mission 02 banner 'and its autosave', which reads as if the checkpoint ended on an autosave. A reader cannot tell why the save differs. Fix: Either state in the why that the autosave is not reliable here and a menu save is taken, or switch to 'autosave' if that is what the game does.

The why and source strings drop console, date, renumbering and earlier-value
notes and keep the reason. 01.02 says its 0 ms leg only turns and that the
sidestep's 1000 ms is not measured yet.
The step numbers refer to the recording, which is not in the repo; each
why keeps what the event does. 01.03's "choose it" names Chikorita.
… none

No run log shows [Autosave] at the end of 01.00-01.02, so they are "unverified"; 01.02 no longer reloads when chained. 01.03 drops "menu".
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