FLYGAME / PLAYER & ENGINEERING REFERENCE

Saving, Continuing & Chest Persistence

FLYGAME WIKI / GAME SYSTEMS / PERSISTENCE · Reviewed November 15, 2026 · Evidence: SOURCE-IMPLEMENTED; FULL CURRENT-ROM SAVE/CONTINUE UNVERIFIED

A Game Boy adventure has to remember what matters

FlyGame's save system is designed to store durable game facts instead of copying all of working memory. That distinction matters when an RPG must reconstruct not only your position, but equipment, quest flags, the world context and temporary effects after you return.

The current save format

The authoritative save-schema generator currently specifies version 9, two 512-byte slots in battery-backed SRAM, and migration from older v7 and v8 formats. Earlier save documents refer to v7 as historical status, not the live schema revision.

  • World position and indoor/outdoor context can be restored.
  • Player progression, current vitals, currency, equipment and inventory have their own durable state.
  • Persistent event flags remember important triggers, puzzles and rewards.
  • Derived combat and weather state should be rebuilt from saved sources rather than blindly copied from volatile RAM.

The title/Continue interface and save components are present in the source, with prior test evidence. The crucial remaining milestone is to verify the present ROM with a private SRAM destination through a complete ordinary-play Save → exit → Continue cycle.

Why two slots?

The save design tracks valid slots and a sequence number so the game can choose the newest valid record. Static tests exercise slot selection and rollover behavior; that is valuable proof of the data contract, not a substitute for loading real saved progress in the emulator.

What about chests?

A chest must not reroll its full contents just because a player took one item, switched contexts or reloaded. The documented contract marks the persistent chest state when its loot session begins. A separate banked return-channel fix was integrated for matching a resumed chest action, with static regression checks.

Current open question: F-024 investigates memory/call-depth safety around debug warp and chest flows. A candidate reduced one compiled local stack frame but has not established full runtime safety. Readers should not assume every chest edge case is accepted.

What must the next playtest prove?

  • Save from the intended game state, not only a debugger-poked fixture.
  • Exit cleanly and restart the exact same ROM in isolated SRAM.
  • Continue with position, progression, equipment, quest/puzzle and chest state intact.
  • Confirm natural weather and Tideheart state are reconstructed as designed.
  • Identify ROM hash, save slot, testing method and any missing or failed step.

Important distinction

An old build having passed a specific save test does not automatically certify a newer changing ROM. The first fully accepted playable slice must carry its own source and save identity.

Sources: current tools/save_schema.py (save v9), Save System Plan (historical v7 architecture, July 2026), Tideheart Weather Runtime Contract (August 2026), November F-024/F-025 integration reports.

Playable Slice Tracker · Weather & Tideheart · Known Issues