-- General

babudb/cpp is a library that implements several database abstractions that can
be combined to build an embedded database. The glue code for building a
non-replicated database with string-type index keys is provided.

babudb is built around the concept of log-structured merge (LSM) trees. A LSM
tree is an index that consists of an immutable persistent base index and one or
several overlays. In combination the base and its overlays yield the current
state of the index. The oldest overlay index can be merged with the base index
in order to generated a more recent base.

babudb manages the consistency of indices with the write-ahead operation log. It
does not define an operation semantics by itself, but merely uses the concept of
log-sequence numbers (LSNs) to associate logged operations with index state.

For more information please refer to the Database class in database.h.

A simple database with several indices that map string keys to string values
can be found in profiles/string_db.h


-- Strength and Weaknesses

babudb is designed as a light-weight database system around an
application-defined log format. If you need full control over the log, babudb
is for you. 

As babudb comes with wrapper classes for a simple mult-index database, it can
also help applications that need a light-weight BSD-licensed embedded database.

Features:
* full control over the operations log
* light-weight and BSD-licensed
* multi-platform, no dependencies

babudb's on-disk formats are still pretty basic and lack several features,
mostly around compression, but also around corruption handling. They will
change.

Therefore don't use babudb if you need:
* a future-proof on-disk format
* thread or process-level concurrency handling
* a mature code base
* a query engine
* large databases on 32 bit systems
* a relational data model


--- Building

babudb/cpp's build system is based on CMake (http://www.cmake.com). With CMake
you can generate build or project files for your favorite build environment
and platform.

CMake exports two targets: a library for linking with your applications and a
unit test executable.

babudb contains yield_platform, an old snapshot of the platform foundation
classes of yield (http://yield.googlecode.com).