|
|
| The SQLite logo as of 2007-12-15 (Photo credit: Wikipedia) |
SQLite was conceived on a battleship by its author,
D. Richard Hipp who was
working for General Dynamics on a program for the U.S. Navy developing software for use on
board guided missile destroyers. The program originally ran on
Hewlett-Packard Unix (HPUX)
and used an
Informix database as the back-end.
In January 2000, Hipp and a colleague discussed the idea of creating a simple embedded
SQL database that would use the GNU DBM B-Tree library (gdbm) as a back-end, one that would
require no installation or administrative support whatsoever. Later, when some free time opened
up, Hipp started work on the project, and in August 2000, SQLite 1.0 was released.
As planned, SQLite 1.0 used gdbm as its storage manager. However, Hipp soon replaced
it with his own B-tree implementation that supported transactions and stored records in key
order. With the first major upgrade in hand, SQLite began a steady evolution, growing in both
features and users. By mid-2001 many projects—both open source and commercial alike—
started to use it. In the years that followed, other members of the open source community
started to write SQLite extensions for their favorite scripting languages and libraries. One by
one, new extensions—an
Open Database Connectivity (ODBC) interface followed by extensions
for Perl, Python, Ruby, Java and other mainstays—fell into place and testified to SQLite’s
wide application and utility.
SQLite began a major upgrade from version 2 to 3 in 2004. Its primary goal was enhanced
internationalization supporting
UTF-8 and UTF-16 text as well as user-defined text-collating
sequences. While 3.0 was originally slated for release in summer 2005,
America Online provided
the necessary funding to see that it was completed by July 2004. Besides internationalization,
version 3 brought many other new features such as a revamped C API, a more compact format
for database files (a 25 percent size reduction), manifest typing,
Binary Large Object (BLOB)
support, 64-bit ROWIDs,
autovacuum, and improved concurrency. In spite of the many new
features, the overall library footprint was still less than 240 kilobytes. Another improvement in
version 3 was a good code cleanup—revisiting and rewriting, or otherwise throwing out extraneous
stuff accumulated in the 2.x series.
