
Version 1.1 - 14 Aug 2006

- First public release
- Added verbose mode to display filenames as they're being hashed
- Added -d mode to make finding similar files in the same directory tree
    both easier and faster. Removes the need for two command lines and
    many extraneous lines of output.
- Added -p mode to improve -d mode. Prints bi-directional matches together
    and omits self matches.
- Added LARGEFILE_SOURCE define to Linux version to allow processing
    of large files. (You never know...) 
- Fixed cosmetic errors in usage message. Updated man page.


Version 1.0 - 31 Mar 2006

- Released internally
- Fixed failure to close files after reading in engine.c
- Fixed routine to read headers of matching hashes on Windows.
- Fixed handling of symbolic links
- Fixed cosmetic bug to display error messages if file open fails
    (e.g. Permission denied, etc)
- Removed quotation marks from the signatures but not the file names.
    Filenames may contain spaces, but signatures may not. Two bytes
    per line adds up when we starting compiling large hash sets.
- Added silent mode, -s. All error messages are suppressed.
- Redirected all error messages to stderr instead of stdout
- Removed duplicate defines at the start of engine.c
- Replaced all references to u32 with C99 standard uint32_t
- Added error checking for memory allocation in main.c:main() and
    engine.c:hash_file()
- Removed useless logical AND of 0xFFFFFFFF from rolling hash update


Version 0.1 - 4 Nov 2005

- Initial release
- This version supports recursion, relative and bare file names, and
    can perform positive matching using a previous output.
