Chapter 18: A-fighting we will go
We are spending time doing various housekeeping activities, code cleanup, minor feature implementation - but the main goal is researching combat and RSD.
During combat, I record what I see on the screen during each round of each encounter, like this: first encounter R1:3(3) slimes, 5(5) small humanoids R2:1(1) bubbly slime, 5(5) orcs R3:5(5) orcs R4:2(2) orcs R5: all monsters killed
Then I upload this and Moose compares the changes in monster party number to his snapshots of RAM. This is allowing Moose to determine exactly where in RAM this data is being held. This is going to free us from relying on OCR (which is unreliable) during combat. We are making good, if slow, progress.
For the blog, please give me a succinct summary of the work done tonight as a bullet list where each bullet is only one sentence
- Fixed maze navigation so interrupted turns and movements roll back cleanly instead of corrupting dead-reckoned position and facing.
- Reworked safe-exit breadcrumb handling so mapped retreat no longer leaves stale fallback directions behind.
- Added full round-by-round combat RAM snapshots to support human-assisted RSD research.
- Fixed combat round detection so human notes like R1, R2, and R3 line up with the correct RAM samples, including surprise rounds.
- Added a full terminal combat RAM snapshot so “all monsters killed” can be correlated with the zero-enemy state.
- Used human combat observations to establish strong candidate patterns for monster identity, group count, and active-count RAM fields.
- Removed the party Treasurer concept completely so Inn exit service no longer depends on having a sixth party member.
- Changed the Inn, Boltac’s, and the Temple so the first action after a character enters is always P)ool gold.
- Squash-merged the night’s 20 commits into main as Chapter 18b.
Great going, Moose! Always a pleasure to work with you.