Files
BFM-decomp/docs
Drew T ce79096547 feat(phase-12): T1 — resident PsyQ footprint survey = NIL (resident is custom engine code)
- psyq_identify over 4.7 (conv47, already ELF) AND 4.0 libs vs the resident
  (MAIN.CD/FILE_010/1.1, --vram-base 0x800CEDF8), code window + whole file:
  NIL footprint — 4.7 libsnd 1/226 (4-ins coincidence), libspu 0/134, libgte
  0/509, libgpu 0/61; 4.0 libsnd 2/163 (VM_DON/VM_DOFF 4-5 ins), libspu/gte 0.
  All hits are the §9.5 short-object coincidental class (<=8 ins).
- tool sanity: 4.0 libsnd vs the EXE snd region = 35/163 (works, version-sensitive)
- Ghidra corroboration: DsMix = {FUN_800d1bf8(); return 1;} custom wrapper (NOT
  stock libsnd; R13 tag refuted); resident makes 61 distinct EXE-range jal calls
  vs 37 internal — it CALLS the EXE's SDK, doesn't embed it
- => the resident is ~143 funcs of custom engine code; nothing to link. The
  DetectPsyQ 4.7 was one coincidental signature. Phase-12 plan revised
  (Drew-approved): drop the library-integration tasks; full substantial harvest
  of the resident engine by hand (REAL, not LINKED). Resident unchanged 8e17e02f.
- docs/psyq-worklist.md: corrected resident section with the byte-evidence +
  regen command; phase-ends/CURRENT_PHASE.md: revised task plan
2026-06-16 11:05:22 -06:00
..