On this page:
6.1 Local Sockets for Postgre  SQL and My  SQL Servers
6.2 Postgre  SQL Database Character Encoding
6.3 Postgre  SQL Authentication
6.4 Postgre  SQL Timestamps and Time Zones
6.5 My  SQL Authentication
6.6 My  SQL CALLing Stored Procedures
6.7 Cassandra Authentication
6.8 SQLite Requirements
6.9 FFI-Based Connections and Concurrency
6.10 ODBC Requirements
6.11 ODBC Status
6.11.1 DB2 ODBC Driver
6.11.2 Oracle ODBC Driver
6.11.3 SQL Server ODBC Driver

6 Notes

This section discusses issues related to specific database systems.

6.1 Local Sockets for PostgreSQL and MySQL Servers

PostgreSQL and MySQL servers are sometimes configured by default to listen only on local sockets (also called “unix domain sockets”). This library provides support for communication over local sockets on Linux and Mac OS. If local socket communication is not available, the server must be reconfigured to listen on a TCP port.

The socket file for a PostgreSQL server is located in the directory specified by the unix_socket_directory variable in the postgresql.conf server configuration file. For example, on Ubuntu 11.04 running PostgreSQL 8.4, the socket directory is /var/run/postgresql and the socket file is /var/run/postgresql/.s.PGSQL.5432. Common socket paths may be searched automatically using the postgresql-guess-socket-path function.

The socket file for a MySQL server is located at the path specified by the socket variable in the my.cnf configuration file. For example, on Ubuntu 11.04 running MySQL 5.1, the socket is located at /var/run/mysqld/mysqld.sock. Common socket paths for MySQL can be searched using the mysql-guess-socket-path function.

6.2 PostgreSQL Database Character Encoding

In most cases, a database’s character encoding is irrelevant, since the connect function always requests translation to Unicode (UTF-8) when creating a connection. If a PostgreSQL database’s character encoding is SQL_ASCII, however, PostgreSQL will not honor the connection encoding; it will instead send untranslated octets, which will cause corrupt data or internal errors in the client connection.

To convert a PostgreSQL database from SQL_ASCII to something sensible, pg_dump the database, recode the dump file (using a utility such as iconv), create a new database with the desired encoding, and pg_restore from the recoded dump file.

6.3 PostgreSQL Authentication

PostgreSQL supports a large variety of authentication mechanisms, controlled by the pg_hba.conf server configuration file. This library currently works with the following authentication methods:
  • plain (and ldap, pam, radius): cleartext password, only if explicitly allowed (see postgresql-connect)

  • md5: MD5-hashed password

  • scram-sha-256: password-based challenge/response protocol

  • peer: only for local sockets

The gss, sspi, krb5, pam, and ldap methods are not supported.

Changed in version 1.2 of package db-lib: Added scram-sha-256 support.

6.4 PostgreSQL Timestamps and Time Zones

PostgreSQL’s timestamp with time zone type is inconsistent with the SQL standard (probably), inconsistent with time with time zone, and potentially confusing to PostgreSQL newcomers.

A time with time zone is essentially a time structure with an additional field storing a time zone offset. In contrast, a timestamp with time zone has no fields beyond those of timestamp. Rather, it indicates that its datetime fields should be interpreted as a UTC time. Thus it represents an absolute point in time, unlike timestamp without time zone, which represents local date and time in some unknown time zone (possibly—hopefully—known the the database designer, but unknown to PostgreSQL).

When a timestamp with time zone is created from a source without time zone information, the session’s TIME ZONE setting is used to adjust the source to UTC time. When the source contains time zone information, it is used to adjust the timestamp to UTC time. In either case, the time zone information is discarded and only the UTC timestamp is stored. When a timestamp with time zone is rendered as text, it is first adjusted to the time zone specified by the TIME ZONE setting (or by AT TIME ZONE) and that offset is included in the rendered text.

This library receives timestamps in binary format, so the time zone adjustment is not applied, nor is the session’s TIME ZONE offset included; thus all sql-timestamp values in a query result have a tz field of 0 (for timestamp with time zone) or #f (for timestamp without time zone). (Previous versions of this library sent and received timestamps as text, so they received timestamps with adjusted time zones.)

6.5 MySQL Authentication

As of version 5.5.7, MySQL supports authentication plugins. The only plugins currently supported by this library are mysql_native_password (the default) and mysql_old_password, which corresponds to the password authentication mechanisms used since version 4.1 and before 4.1, respectively.

6.6 MySQL CALLing Stored Procedures

MySQL CALL statements can be executed only if they return at most one result set and contain no OUT or INOUT parameters.

6.7 Cassandra Authentication

Cassandra, like MySQL, supports authentication plugins. The only plugins currently supported by this library are AllowAllAuthenticator and PasswordAuthenticator.

6.8 SQLite Requirements

SQLite support requires the appropriate native library.

6.9 FFI-Based Connections and Concurrency

Wire-based connections communicate using ports, which do not cause other Racket threads to block. In contrast, an FFI call causes all Racket threads to block until it completes, so FFI-based connections can degrade the interactivity of a Racket program, particularly if long-running queries are performed using the connection. This problem can be avoided by creating the FFI-based connection in a separate place using the #:use-place keyword argument. Such a connection will not block all Racket threads during queries; the disadvantage is the cost of creating and communicating with a separate place.

6.10 ODBC Requirements

ODBC requires the appropriate driver manager native library as well as driver native libraries for each database system you want use ODBC to connect to.

In addition, you must install the appropriate ODBC Drivers and configure Data Sources. Refer to the ODBC documentation for the specific database system for more information.

6.11 ODBC Status

ODBC support is experimental. The behavior of ODBC connections can vary widely depending on the driver in use and even the configuration of a particular data source.

The following sections describe the configurations that this library has been tested with.

Reports of success or failure on other platforms or with other drivers would be appreciated.

6.11.1 DB2 ODBC Driver

The driver from IBM DB2 Express-C v9.7 has been tested on Ubuntu 11.04 (32-bit only).

For a typical installation where the instance resides at /home/db2inst1, set the following option in the Driver configuration: Driver = /home/db2inst1/sqllib/lib32/libdb2.so. (The path would presumably be different for a 64-bit installation.)

The DB2 driver does not seem to accept a separate argument for the database to connect to; it must be the same as the Data Source name.

6.11.2 Oracle ODBC Driver

The driver from Oracle Database 10g Release 2 Express Edition has been tested on Ubuntu 11.04 (32-bit only).

It seems the ORACLE_HOME and LD_LIBRARY_PATH environment variables must be set according to the oracle_env.{csh,sh} script for the driver to work.

Columns of type TIME can cause a memory error (ie, Racket crashes). This seems to be due to a bug in Oracle’s ODBC driver, but I do not yet have a workaround.

6.11.3 SQL Server ODBC Driver

Basic SQL Server support has been verified on Windows, but the automated test suite has not yet been adapted and run.

The “SQL Server” driver refuses to accept NUMERIC or DECIMAL parameters, producing the error “Invalid precision value (SQLSTATE: HY104).” If possible, use the “Native SQL Server” driver instead.