postgresql/src/interfaces/jdbc

This is a simple readme describing how to compile and use the jdbc driver.

This file was amended on May 2 2000 to document the changes made in the 7.0
release.

---------------------------------------------------------------------------

This isn't a guide on how to use JDBC - for that refer to Javasoft's web site:

	http://www.javasoft.com

For problems with this driver, then refer to the postgres-interfaces email
list:

	http://www.postgresql.org

---------------------------------------------------------------------------

COMPILING

There are actually two versions of the driver. One for the JDBC1.2 specification, and one for the JDBC2 specification. To compile the driver, you need to select the correct one.

If you have JDK1.1.x you need to type: make jdbc1

If you have JDK1.2 or JDK1.3, you need to type: make jdbc2

This will compile the driver, and build a .jar file (Java ARchive) called
postgresql.jar

That jar file will contain the driver for _your_ version of the JDK.

REMEMBER: Once you have compiled the driver, it will work on ALL platforms
that support that version of the API. You don't need to build it for each
platform.

That means you don't have to compile it on every platform. Believe me, I
still hear from people who ask me "I've compiled it ok under Solaris, but it
won't compile under Linux" - there's no difference.

PS: When you run make, don't worry if you see more than one or two calls to
    javac. This is normal, because the driver dynamically loads classes, and
    the Makefile ensures everything gets compiled.

I advise you don't try running javac outside of make. You may miss something.

Possible problems

You may see a message similar to:

postgresql/Driver.java:87: interface java.sql.Connection is an interface. It can't be instantiated.
    return new Connection (host(), port(), props, database(), url, this);

This is caused by not having the current directory in your CLASSPATH. Under
Linux/Solaris, unset the CLASSPATH environment variable, and rerun make.

If you are still having problems, I keep a copy of the driver (for different
versions of the backend) on my web site http://www.retep.org.uk/postgres/

---------------------------------------------------------------------------

INSTALLING THE DRIVER

To install the driver, the .class files have to be in the classpath. To do
this, copy the postgres.jar file into a directory, and add it to the classpath.

ie: under LINUX/SOLARIS (the example here is my linux box):

	export CLASSPATH=.:/usr/local/lib/postgresql.jar

---------------------------------------------------------------------------

USING THE DRIVER

To use the driver, you must introduce it to JDBC. Again, there's two ways
of doing this:

1: Hardcoded.

   This method hardcodes your driver into your application/applet. You
   introduce the driver using the following snippet of code:

	try {
	  Class.forName("org.postgresql.Driver");
	} catch(Exception e) {
	  // your error handling code goes here
	}

   Remember, this method restricts your code to just the postgresql database.
   However, this is how most people load the driver.

2: Parameters

   This method specifies the driver from the command line. When running the
   application, you specify the driver using the option:

	-Djdbc.drivers=org.postgresql.Driver

   eg: This is an example of running one of my other projects with the driver:

	java -Djdbc.drivers=org.postgresql.Driver uk.org.retep.finder.Main

   note: This method only works with Applications (not for Applets).
	 However, the application is not tied to one driver, so if you needed
	 to switch databases (why I don't know ;-) ), you don't need to
	 recompile the application (as long as you havent hardcoded the url's).

---------------------------------------------------------------------------

JDBC URL syntax

The driver recognises JDBC URL's of the form:

	jdbc:postgresql:database

	jdbc:postgresql://host/database

	jdbc:postgresql://host:port/database

Also, you can supply both username and passwords as arguments, by appending
them to the URL. eg:

	jdbc:postgresql:database?user=me
	jdbc:postgresql:database?user=me&password=mypass

---------------------------------------------------------------------------

That's the basics related to this driver. You'll need to read the JDBC Docs
on how to use it. However, there are some examples included in the example
directory. To build, type: make examples

To run them, they follow the same syntax. For example, the basic example shows
how to insert data, and perform queries:

	java example.basic jdbc:postgresql:test user password

---------------------------------------------------------------------------

POSTGRESQL SPECIFICS
--------------------

Large Objects:

A "feature" of PostgreSQL is that access to LargeObjects is only permitted
within a Transaction. Because of this, any use of LargeObjects (also known
as Blobs) requires that the Connection.setAutoCommit() method be called
disabling the autocommit feature.

For example:

	Connection db;			// open the connection here
	db.setAutoCommit(false);	// Turn off AutoCommit

			------------------

Large Object API

Most of the time, you can use the getBytes()/setBytes() methods to read and
write small Large Objects. However, PostgreSQL's own internal api's are
available. These allow you to access the object as if it was a file.

The first thing you need to do is to open the LargeObjectManager. This class
handles the opening of existing objects, and creating new ones. To do this,
you use the following line of code:

	LargeObjectManager lobj;
	lobj = ((org.postgresql.Connection)db).getLargeObjectAPI();

where db is a reference to an open Connection object.

Once that is done, you can use the API for the lifetime of that Connection.

To create an object, you call the create() method. This takes an argument
with the file modes you intend to use. The following line is normally
sufficient:

       int oid = lobj.create(LargeObjectManager.READ|LargeObjectManager.WRITE);

Here, lobj is the LargeObjectManager we have opened earlier, and oid is the
Large Object's oid in the database.

To open an existing object, you use the open() method. This takes an oid, and
the file permissions. It then returns a LargeObject object.

	LargeObject obj = lobj.open(oid,LargeObjectManager.WRITE);

Once the LargeObject is open, you can call methods to read, write, seek etc.
Here's the supported methods:

	int oid = obj.getOID();			Return the objects oid
	obj.close();				Close the object
	byte data[] = obj.read(int len);	Read len bytes
	onj.read(byte data[],int off,int len);	Read into data[off] len bytes
	obj.write(byte data[]);			Write the array data
	obj.write(byte data[],int off,int len);	Write len bytes from data[off]
	obj.seek(int pos,int ref);		As fseek in C.
	obj.seek(int pos);			Move to pos (from the begining)
	int pos = obj.tell();			Returns the current position
	int size = obj.size();			Returns the objects size

Caveat: If you commit(), rollback() a transaction, or turn on autocommit whilst
an object is open PostgreSQL will close it. You will need to reopen the object
before using it again. Using the existing LargeObject will cause an
SQLException to be thrown.

			------------------

Date datatype:

The driver now issues the "show datestyle;" query when it first connects, so
any call to ResultSet.getDate() how returns the correct date.

One caveat though: if you change the datestyle from within JDBC, you must also
issue the "show datestyle" query. Without this, the driver will not know of
the change.

ie:
	Statement s = db.createStatement();
	...
	s.executeUpdate("set datestyle='european'");
	s.executeUpdate("show datestyle");
	..
	s.close();

Please note: This may change later, so that the driver uses the same format
internally (similar to how the ODBC driver works).

			------------------

JDBC supports database specific data types using the getObject() call. The
following types have their own Java equivalents supplied by the driver:

	box, circle, line, lseg, path, point, polygon

When using the getObject() method on a resultset, it returns a PG_Object,
which holds the postgres type, and its value. This object also supports
methods to retrive these types.

	Eg: column 3 contains a point, and rs is the ResultSet:

	PG_Object o = (PG_Object)rs.getObject(3);
	PGpoint p = o.getPoint();
	System.out.println("point returned x="+p.x+", y="+p.y);

Also, when using these classes, their toString() methods return the correct
syntax for writing these to the database.

---------------------------------------------------------------------------

Peter T Mount, December 29 1998
home email: pmount@retep.org.uk		http://www.retep.org.uk
work email: petermount@it.maidstone.gov.uk or peter@taer.maidstone.gov.uk

PS: Please use the home email whenever possible. If you must contact me at work
then please cc my home one at the same time.