HOLYC STATIC VERIFICATION PASS — 2026-07-11
===========================================
Cannot run TempleOS's HolyC compiler in this Linux container, so did the
strongest static check possible: every function call in HDGLAgent.HC and
PaneBus.HC checked against its ACTUAL signature in the TempleOS-master source.

SIGNATURES VERIFIED (file:line in TempleOS-master):
  FileRead   Kernel/BlkDev/DskFile.HC:1   U8 *FileRead(U8*,I64*_size=NULL,...)
  FileWrite  Kernel/BlkDev/DskFile.HC:79  I64 FileWrite(U8*,U8*fbuf,I64 size,...)
  FileFind   Kernel/KExts.HC:56           Bool FileFind(U8*,CDirEntry*=NULL,...)
  Del        Kernel/KExts.HC:31           I64 Del(U8*files_find_mask,...)
  MStrPrint  Kernel/KExts.HC:78           U8 *MStrPrint(U8*fmt,...)
  ExePutS    Compiler/CMain.HC:571        I64 ExePutS(U8*buf,...)   [self-mod JIT]
  Spawn      Kernel/KTask.HC:242          CTask *Spawn(fp,data,name,...)
  RandU64    Kernel/KMathB.HC:127         U64 RandU64()             [LCG^GetTSC]
  Seed       Kernel/KMathB.HC:147         I64 Seed(I64 seed=0,...)
  try/catch/PutExcept — native HolyC (used in Kernel/KEnd.HC etc.)  ✓

BUG FOUND + FIXED (PaneBus.HC):
  Original cleared the bus with FileWrite(path,"",0) then looped on
  FileFind(path). An empty file still EXISTS -> FileFind stays TRUE ->
  the loop would re-trigger on the same message forever.
  FIX: consume-then-Del(path). Del makes FileFind go FALSE. Also guard the
  ExePutS on sz>0 (from FileRead's out-param) so an empty read never compiles.
  Sidecar side checked too: it guards on [ -s ] (size>0), so its truncate is
  already safe — no symmetric bug.

STILL CANNOT VERIFY (needs real TempleOS):
  - that the HolyC COMPILES clean (the (I64) cast form, F64->U8 truncation,
    fixed-point #define may need tweaks — flagged in STATUS.txt).
  - ExePutS re-entrancy inside a Spawn'd task at runtime.
  These are the honest remaining unknowns; every symbol + arg-order is right.
