Movies, scenarios and the first routes (drive the game from its state) #19
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Brief §8.2-8.3. Needed by #8 (RAM-executed code on menu/world/ending) and #11 (game behaviour with ports on).
docs/): header with ROM CRC32 and version, power-on seed (decision D2), start point (power-on, or SRAM file hash), frame count, LakeSnes revision; then pad state per frame.ff6cli --movie <file>plays one;--record <file>writes the movie of a run (from--inputor the drive server).workdocs/reuse-audit.md§4): human-writable JSON, coarse timeline in seconds (wait,hold ... for), compiles to a movie.tests/routes/, Python overff6cli --serve): a small behaviour library reading the RAM map by name (wait for a dialogue window and confirm, detect battle start/end, pick a battle command by reading the cursor, open the field menu), checkpoints with in-memory save/load and retry. First routes:title_to_first_battle_win: power-on -> New Game -> opening -> first Narshe battle won -> back on the field; semantic checkpoint at the end (party HP, field map id, event flags).field_menu_tour: open the main menu on the field and visit each sub-menu, then close.Each route records its movie (
tests/movies/, no ROM data) and the hashes of build/ROM/script.slow, skipped without ROM), US and JP.Starting: branch
routes, delegated to a fresh agent, reviewed and merged by the main session.Merged (PR #20). Main-session review: release
ctest -LE slow13/13; independent replay oftests/movies/title_to_first_battle_win.us_v10.ff6mwith--port all=shadow: movie match (picture, WRAM, SRAM hashes) andbattle/Rand17/17 over the whole won battle; screenshots checked (Magitek beam during the battle at frame 9200, party back on the Narshe field at 10253). Agent run: 24/24 release tests incl. slow, US and JP.