README

	FFDB File Implementation on top of file hash
	     	   Feb. 19, 2009

FFDB file is a part of FFDB. It can be used without other packages such as
ADAT. The purpose of this package is allow applications to store pair of
data into a hash DB database. A pair of data contains a key and a datum.
The key has to be a subclass of DBKey and the datum has to be a subclass of
DBData. Applications should use ConfDataStoreDB class to store keys and 
associated data.  The ConfDataStoreDB is a template class. It should be
intantiated using one DBKey subclass and one DBData subclass. This class
assumes that the a key template paramter has following functions defined:

writeObject(),
readObject(),
sizeOfObject(),
objectBuffer(),
serialID()
hasHashFunc()
hasCompareFunc()

In addition, two static functions with names of hash and compare. These two 
functions serves as database hash and compare functions if applications
have better idea how to handle their keys. If a DBKey subclass decides to
use Berkeley DB default hash or compare functions, the class should have 
hasHashFunc() or hasCompareFunc() return 0 respectively and should have
empty hash and compare functions. Finally, a DBKey subclass should also
implement '=' and '==' operators which are used inside the DBCursor class.

Fortunately, a subclass of DBData has fewer functions to implement. The
functions are:

writeObject(),
readObject(),
sizeOfObject(),
objectBuffer(),
serialID()

These two classes assumes there is an internal buffer converting C++ object
into a binary form which is network byte ordered.


See INSTALL file for installtion instructions.

Send bugs and comments to chen@jlab.org