This distribution contains Tie::DBI and Tie::RDBM, two modules that allow you to tie associative arrays to relational databases using the DBI library. The hash is tied to a table in a local or networked database. Reading from the hash retrieves values from the datavbase. Storing into the hash updates the database (if you have sufficient privileges). Tie::DBI implements a fully general binding between the hash and the table, allowing you to access fields in this manner: $h{strawberries}->{quantity} = 228; Tie::RDBM implements a more limited binding which gives you access to only one column of the table. However it is faster, can create and destroy tables on the fly, and provides an interface whereby you can store and retrieve complex Perl objects from the database table. In addition to this package, you will need the DBI database interface module, plus the DBD driver for the database of your choice. To be able to store and retrieve non-scalar values via the Tie::RDBM interface you will also need Storable.pm These are all available at a CPAN near you: http://www.perl.com/CPAN/ To install: 1. perl Makefile.PL 2. make 3. make test [optional] 4. make install To read the documentation: perldoc Tie::DBI perldoc Tie::RDBM "make test" requires you to have a database named "test" up and running on the local machine. The database must allow table add and drop privileges for the current user. The test script will search for an installed DBD driver to use. You can override its choice by specifying DRIVER= on the command line. For example: make test DRIVER=mysql If you need to specify a different database, or need to provide a username and password to log in, you can provide these values as DB, USER and PASS, respectively, as in: make test DRIVER=Oracle DB=DEMO USER=fred PASS=xyzzy And some drivers, such as Pg, want funny database names, as in: make test DRIVER=Pg DB=dbname=test USER=fred PASS=xyzzy Databases differ in their capabilities, and some can't support binary data types (or support them in a difficult, non-standard way). Such databases include mSQL, Postgres and Informix. On those databases, it is not possible to store arbitrary Perl data structures into Tie::RDBM hashes. For this reason Tie::RDBM tests #13 and #14 will be skipped. The test will fail if you do not have Storable.pm installed on your system. The Postgres driver breaks when you try to set a field to null by passing it an under variable. This causes Tie::DBI test #29 to fail. In addition, there is vast heterogeneity among DBD drivers which renders this interface inoperative for certain databases. It has been tested and is known to work on the following databases: 1) DBD::mysql 2) DBD::Oracle 3) DBD::mSQL (skips Tie::RDB tests 13-14) 4) DBD::CSV (skips Tie::DBI test #29) 5) DBD::Postgres (skips Tie::RDBM tests 13-14 and flunks Tie::DBI test #29) 6) DBD::Informix (skips Tie::RDBM tests 13-14) 7) DBD::Sybase (partial, croaks on each()) 8) DBD::Solid 9) DBD::ODBC (on mysql ODBC driver) Microsoft SQL server works partially with Tie::DBI (everything but "each") and not at all with Tie::RDBM. This is because SQL server doesn't allow there to be multiple pending queries on the database. Let me know if you can get it to run with true Sybase servers, and any other drivers you try. For more information, see the perldoc documentation. Also be sure to check out Gerald Richter's DBIx::Recordset, which puts a tied array interface on the results of arbitrary SQL expressions and is optimized for use in embperl. The latest version of this module is available on CPAN. References: The DBI home page: http://www.hermetica.com/technologia/perl/DBI Gerald Richter's DBIx::Recordset: http://perl.apache.org/embperl.html. Lincoln Stein lstein@cshl.org