-----
v1.39 (PUBLIC PASCAL RELEASE)
-----
  * TRACER.DOC updated.
  * Fixed -F option; address was not being set to $7000, which is the
    default address for FFE trainers.

  BE AWAITING THE NEW 386-ASM (PROTECTED MODE) RELEASE OF TRACER! Yes,
  it is in the works, and is coming along quite nicely. It's looking
  good, and the speed is phenominal. Keep your eyes open!

-----
v1.38 (PUBLIC PASCAL RELEASE)
-----
  * TRACER.DOC updated (severely).
  * I don't think I'll be able to stress this enough, but:

         ALL COMMAND-LINE OPTIONS HAVE BEEN CHANGED!
             ALL OPTIONS ARE NOW CASE SENSITIVE!

  * Added -F option; disassembles NES trainer (usually FFE).
  * Minor optimizations were performed; I found a few places in some
    assembly subroutines which were unoptimized.

-----
v1.36 (PUBLIC PASCAL RELEASE)
-----
  * TRACER.DOC updated.
  * Added -x option; enables use of file-offset specification. This option
    is flakey, and should always be used as the *LAST* command line option,
    i.e.:
      TRACER -k -n -o -x 4096 LOLO.NES
    I don't know what will happen if it's used before.
    Also, be aware of the '4096' parm; it specifies the file offset in dec-
    imal.

-----
v1.35 (PUBLIC PASCAL RELEASE)
-----
  * TRACER.DOC updated.
  * Added -1 option; force 16K NES image.
  * Bugs fixed:
    * The NES uses a 6502 opcode set, not a 65C02 opcode set. Therefore,
      some opcodes were implemented which shouldn't have been.
  * Optimizations:
    * Branch instructions should now disassemble faster, and there is
      no chance of failure :-).

-----
v1.33 (PUBLIC RELEASE)
-----
  * TRACER.DOC updated.
  * -r option removed, due to non-use.
  * Added -h option; force HiROM disassembly. Therefore, TRaCER now
    supports HiROM disassembly.

-----
v1.31 (PUBLIC RELEASE)
-----
  * Bugs fixed:
    * Opcodes with no operands which are cut off prematurely due to the
      EOF marker were not being translated correctly, the result showing
      up as:
               00/92FC: 00 --        BRK (garbage chars here)

-----
v1.30 (PUBLIC RELEASE)
-----
  * TRACER.DOC updated.
  * Added -b option; disassembles from NMI/VBL vector.
  * Bugs fixed:
    * -v was actually disassembling from the NMI/VBL vector, not the reset
      vector. Oops.

-----
v1.28 (PUBLIC RELEASE)
-----
  * TRACER.DOC updated.
  * Added -v option; disassembles from reset vector.
  * I have removed support for doing redirects under DOS. Stuff like:
      TRACER FILE.BIN > FILENAME
    ...will no longer work. Please use the -o option instead of using the
    DOS '>' redirector. This can also be considered an optimization, as the
    default output device has now changed from a C-style STDOUT to the
    video driver.
  * Optimizations:
    * I have increased the speed of the hex-to-ascii routines once again.

-----
v1.26 (PUBLIC RELEASE)
-----
  * Goodbye protected-mode. Seems Borland Pascal enjoys requiring weird
    run-time-modules, such as their RTM.EXE, when compiling in protected
    mode. I didn't realize this until Dan Boris informed me of the error.
    I feel like a complete git at this point, so i'll stick with the stand-
    ard DOS real mode from now on. :-)

-----
v1.25 (PUBLIC RELEASE)
-----
  * Bugs fixed:
    * Oops. I accidentally removed the Seek() line, therefore -k and -s
      were not working. My bad.

-----
v1.24 (PUBLIC RELEASE)
-----
  * TRACER.DOC updated.
  * SAMPLE.DIS updated.
  * TRaCER now has a web page. Be sure to check out:
      http://www.winternet.com/~shadow/
    And follow the link to find TRaCER hosted off of my own LAN.
  * Optimizations:
    * TRaCER is now a protected-mode application (yes, despite it's small
      size). It now uses the WinAPI library under DOS, providing the ability
      to address all memory available. This explains the size increase. Be
      prepared to experience labelling in the near future.
      This also means it now requires a 386 or higher to be executed. I doubt
      anyone uses TRaCER on a 286, so it's for a just cause.
    * The hex-to-ascii byte/word routines have been re-written in pure
      assembly. As TRaCER uses these functions constantly, a speed increase
      is definitely due.

-----
v1.22 (PUBLIC RELEASE)
-----
  * Bugs fixed:
    * For the zillionth time, a bug was found in the branch instructions.
      The previous address/operand was not being cleared, and therefore was
      causing emotional distress when it came to other operands.
        Wrong: 00/80C9: 54 CF 88     ???
               00/80CC: FC 00 80     JMP $8000
        Right: 00/80C9: 54           ???
               00/80CA: CF           ???
               00/80CB: 88           DEY
               00/80CC: FC 00 80     JMP $8000

-----
v1.21 (PUBLIC RELEASE)
-----
  * Optimizations:
    * Main loop has been sped up by removing dual 'IF's, replacing them
      with 'IF ((x) and (x))'.
  * Bugs fixed:
    * When using -n, opcodes which are not 65c02-compatible were still using
      the actual size of the opcode even in 65c816 mode.
        Wrong: 00/80C9: 54 CF 88     ???
               00/80CC: FC 00 80     JMP $8000
        Right: 00/80C9: 54           ???
               00/80CA: CF           ???
               00/80CB: 88           DEY
               00/80CC: FC 00 80     JMP $8000
    * When using -n, opcodes which are not 65c02-compatible were not calc-
      ulating the offset correctly. This relates directly to the aforemen-
      tioned bug.

-----
v1.20 (PUBLIC RELEASE)
-----
  * Bugs fixed:
    * When using -n, program counter was not being incremented correctly
      if a bad opcode was found, causing the rest of the disassembly to
      be 100% inaccurate.
        Wrong: 00/8016: FF A2 34 12  ???
               00/8017: A7 00        ???
        Right: 00/8016: FF A2 34 12  ???
               00/801A: A7 00        ???

-----
v1.19 (PUBLIC RELEASE)
-----
  * Bugs fixed:
    * Branch operand/address translation was hosed (AGAIN). This time I
      hope I have finally fixed the problem.
    * Stack-relative indexed Y was being displayed wrong.
        Wrong: LDA ($00,Y),S
        Right: LDA ($00,S),Y
    * Jump indirect indexed X was being displayed wrong.
        Wrong: JMP ($1234),X
        Right: JMP ($1234,X)
  * Optimizations:
     * Combine NES & SNES translation routines to provide a smaller and
       more efficient binary.

-----
v1.15 (PUBLIC RELEASE)
-----
  * Optimizations:
     * Address search-and-replace routine has now been replaced with an
       assembly version of the "world-famous" Toad Hall character search.

-----
v1.14 (PUBLIC RELEASE)
-----
  * Added -k option; skips 16-byte iNES header.

-----
v1.13 (PUBLIC RELEASE)
-----
  * Bugs fixed:
     * Minor assembly modifications done to the translation routines,
       primarily the NES translation.
  * TRaCER, from now on, will be officially released by Damaged Cybernetics,
    not OldSkoOL.

-----
v1.12 (BETA RELEASE)
-----
  * Added -n option; supports NES (65c02) opcode output. I'm sure this
    feature has bugs -- if you find any, EMail me!
  * Re-added -t option.
  * Bugs fixed:
     * MVN/MVP were being displayed wrong.
  * Optimizations:
     * Address search-and-replace routine translated to pure assembly.
       Previous versions used BP's Pos() routine, which was hellishly slow.

-----
v1.11 (BETA RELEASE)
-----
  * Added disassembler comparison chart to TRACER.DOC.
  * Added -o option; redirect output to file OUTPUT.DIS.
  * Bugs fixed:
     * Branch/jump addresses were totally wrong; initialization routine
       wasn't 100% correct.
  * Optimizations:
     * Lookup table modified by 1 byte (again). Knocked off another 256
       bytes.
     * Translation routines have been optimized, though are still in
       Pascal -- really should be written in assembly.
     * Minor modifications to actual compile-time options in BP, resulting
       in 512 bytes being knocked off the binary size.

-----
v1.10 (PUBLIC RELEASE)
-----
  * Bugs fixed:
     * 'PER' was not being disassembled correctly.
     * Opcodes such as 'RTS' and 'NOP', which didn't have operands, were
       being displayed wrong.
  * Optimizations:
     * More minor assembly conversions.
  * TRACER.DOC modified to reflect command-line changes and etc...
  * SAMPLE.DIS included as an example disassembly.

-----
v1.09 (BETA RELEASE)
-----
  * Bugs fixed:
     * Disassembled 'MVN' and 'MVP' opcodes were stating the banks in the
       reverse order.
  * Optimizations:
     * REP/SEP checking routine translated to pure assembly.

-----
v1.08 (BETA RELEASE)
-----
  * -t option removed; useless without label support.
  * -r option added; now supports "real" 65816 addressing..
  * Optimizations:
     * Lookup table modified by 1 byte; may seem like nothing, but when you
       have 256 opcodes, it knocks off 256 bytes from the binary.
     * Small optimizations to translations routines made.
     * Variable initialization done via assembly.

-----
v1.07 (BETA RELEASE)
-----
  * Bugs fixed:
     * Minor typos, such as 24-bit addresses being shown as '$bbxxxx' in
       comparison to '$bbxxxx'.
  * Optimizations:
     * Completely re-wrote hex-to-opcode/operand translation routine.
     * Small size modifications made, making the actual executable smaller.
     * Removed use of the DOS include library.

-----
v1.06 (PUBLIC RELEASE)
-----
  * -p option removed, due to non-use.
  * 65c02 compatible flags added to opcode table. NES compatibility should
    be enabled by v1.20.
  * Bugs fixed:
     * End-of-file bug fixed; '--' and 'EOF' markers now shown.
     * REP/SEP following was beyond broken; it now works correctly.
  * Optimizations:
     * Disk access loop sped up by ~25%.
     * Small loops converted to assembly.
     * Opcode table optimized slightly.

-----
v1.05 (PUBLIC RELEASE)
-----
  * Many typos fixed (especially in opcode list).
  * Bugs fixed:
     * When not using -s, seek() was seeking to the second byte of the
       file, not the first.
     * Certain opcodes were using wrong addressing modes; mainly, BIT,
       AND, and ROL.
  * Optimizations:
     * More code converted to pure assembly.
     * Internal variables decreased in size (saving about 32 bytes).
     * REP/SEP following has been sped up.

-----
v1.02 (PUBLIC RELEASE)
-----
  * CHANGES file added.
  * Small optimizations made.

-----
v1.01 (PUBLIC RELEASE)
-----
  * Bugs were found in the branch statements; I was not taking into account
    the actual opcode + operand when calculating the new address to print.
    Therefore, branch statement addresses were 2 bytes off, and long
    branch statement addresses were 3 bytes off.

-----
v1.00 (PUBLIC RELEASE)
-----
