Version 0.0.5
 * Add binary tree indexing library
 * New cleaner mem pool code
 * New cleaner, faster file buffering code
 * Rename to skunkdb

Version 0.0.4
 * Use FNV1-A hash
 * Remove skunk_hash.h
 * Tidy up skunk_query
 * Don't accept zero-length keys
 * Autoconfiscate
 * Nicer API
 * Rewrite database creation, much more memory efficient
 * sdb_make uses stdin/stdout
 * sdb_dump uses stdin
 * Drop all extraneous crap

Version 0.0.3
 * Experimental python bindings
 * Tidy up skunk_write a bit
 * Optimisie benchmark mode
 * Tidy up sdb_dump buffer allocation code

Version 0.0.2
 * Less breathing space. makes for smaller dbs, just as fast
 * Doubled the speed of sdb_dump
 * Create skunk library (skunk.a)
 * sdb_dump much more robust
 * better error reporting in sdb_query
 * sdb_make MUCH MUCH more robust
 * tidy up make_write()
 * Added sdb_tar application
 * memcmp() keys before returning OK

Version 0.0.1
 * Moved key and data lengths in to hash table
 * Rewrote sdb_dump for the change
 * Changed to Fowler/Noll/Vo hash
 * Use addition rather than multiplication for composite hash
 * Give breathing space to hash tables to reduce chaining
 * Validate data from DB in query