New stuff in QDASM v1.00
  - FFREE opcode added

New stuff in QDASM v0.95
  - DO_Prefix is a new extern in QDASM that indicates if the output
    is just a prefix (segment reg overrides, REPx, LOCK, etc.)
  - The output can now be 'really' compiled by an assembler.
  - If a segment Reg is not used in the instruction then it's outputed
    on it's own line.  I've also took into account data/addr size
    overrides which may appear as 'DB 6xH' for now (iff they are
    before the unused seg override).  A segment override is only used
    if the instruction contains a MR (mem-reference).  So 'lods',etc.
    do not have a MR so the seg override will appear on its own
    line before these instructions.  
  - invalid opcode now produces 'DB xxH ;-=invalid opcode=-' so it
    could be assembled if you want.

New stuff in QDASM v0.94
  - fixed 8bits extending.  Was only printing 8bits when it could be
    16 or 32 bits.
  - added support for wierd opcodes where 8bits are sign extended to 8bits.
    No assembler/compiler will ever generate them but if DEBUG.EXE
    can do them then so will QDASM.

New stuff in QDASM v0.93
  - added a space after 'bptr' etc. (minor fix)

New stuff in QDASM v0.92
  - the registers used in mem-ref calculations must be given in
    RERVERSED PUSHAD format.  This is not a bug in DASM but this bad info
    cause a bug in QDebug.  So you must build yourself a Reversed PUSHAD
    table and point ECX to it before calling DecodeOpcode()
  - now supports program base (EBX) so that relative JMPs will be
    usefull in non-zero based apps (COM files, PMODE16 apps, PMODE32
    apps that do not use a Zero Base, etc.)
    Simply load EBX with your program's Base each time calling DecodeOpcode()
    Before this it was always assumed the Base was Zero.
    You may now also implement Virtual Addressing:
      What is it?
        Virtual Address is forcing the disassembler to print
        addresses which are what you need.
        For example, in DASMCOM (the example provided) the COM file is loaded
        well above 100h (thru malloc()) so when Q-DASM is used to disassemble
        the file it normally appears as though the COM file does not start
        at 100h which can look very awkward.
        But thru the use of 'EBX' you can force it to start at 100h.
      Example:
        mov eax,malloc_ed_buffer
        neg eax
        add eax,100h
        mov prg_base,eax
      Now simple load EBX with prg_base each time you use DecodeOpcode()
      and the disassemble code will look as though it is starting at 100h
      (even if in reality you have it loaded well above 1MB).  DASMCOM
      now uses this to make the disassemble look normal.
    Of course you don't need to use this if your program is loaded in
    memory where it is support to be (then simply load EBX with the program
    base.)
    And if you are using Zero Based Selectors (PMODE32) then simply
    load EBX with Zero each time you use DecodeOpcode() now.

New stuff in QDASM v0.91
  - many 32bit addressings were not properly decoded
  - some relative jmps where not properly decoded
  - FLAGS.INC was not correct so some Jx comparisons were incorrect
  - the JLE and JG where not properly comparied

New stuff in QDASM v0.90
  - 1st release, based on Dazmit v1.03 by fluff@geocities.com
