Package org.apache.derby.jdbc

JDBC driver and data source implementations.

See:
Description

Class Summary

ClientBaseDataSource	Base class for client-side DataSource implementations.
ClientConnectionPoolDataSource	ClientConnectionPoolDataSource is a factory for PooledConnection objects.
ClientDataSource	ClientDataSource is a simple data source implementation that can be used for establishing connections in a non-pooling, non-distributed environment.
ClientDriver
ClientXADataSource	This is Derby's network XADataSource for use with JDBC3.0 and JDBC2.0.
EmbeddedConnectionPoolDataSource	EmbeddedConnectionPoolDataSource is Derby's ConnectionPoolDataSource implementation for the JDBC3.0 environment.
EmbeddedDataSource	EmbeddedDataSource is Derby's DataSource implementation for JDBC3.0.
EmbeddedDriver	The embedded JDBC driver (Type 4) for Derby.
EmbeddedSimpleDataSource	EmbeddedSimpleDataSource is Derby's DataSource implementation for J2ME/CDC/Foundation.
EmbeddedXADataSource	EmbeddedXADataSource is Derby's XADataSource implementation for JDBC3.0.

Package org.apache.derby.jdbc Description

JDBC driver and data source implementations.

Client/Remote driver and data sources, used to connect to the network server

• ClientDriver - Type 4 JDBC driver
• ClientDataSource - Data source
• ClientConnectionPoolDataSource - Connection pooling data source
• ClientXADataSource - XA data source

Embedded driver and data sources, used when the database engine is embedded with the application.

• EmbeddedDriver - Type 4 JDBC driver
• EmbeddedDataSource, EmbeddedDataSource40 - Data sources for J2SE and J2EE
• EmbeddedSimpleDataSource - Data source for J2ME/CDC/Foundation and J2SE
• EmbeddedConnectionPoolSource, EmbeddedConnectionPoolSource40 - Connection pooling data source
• EmbeddedXASource, EmbeddedXASource40 - XA data sources

Derby's JDBC api is defined by its entry point classes, the drivers and data source implementations and the standard JDBC api definitions of the java.sql and javax.sql classes. Derby does not provide non-standard extensions to standard JDBC classes such as Connection, to encourage portable JDBC applications.

Derby's implementation notes for JDBC

The JDBC specification is sometimes unclear or contradictory. This section covers how Derby implements some JDBC features and methods where the specification might be unclear. The JDBC specicification is in theory a single document however clarifications or additional information can also be found in the Javadoc for java.sql and javax.sql, and in the offical JDBC tutorial book.

• [JDBC3] Downloadable from http://java.sun.com/products/jdbc/index.jsp
• [JAVADOC14] downloadable from http://java.sun.com/products/jdbc/index.jsp
• [TUTORIAL3] JDBC API Tutorial and Reference, Third Edition. ISBN 0321173848 http://java.sun.com/developer/Books/jdbc/Fisher/index.html

JDBC Implementation notes by JDBC class
java.sql.Blob	java.sql.Clob	java.sql.Connection
java.sql.PreparedStatement	java.sql.ResultSet
JDBC Implementation notes by SQL type
SQL DATE and JDBC	SQL TIME and JDBC	SQL TIMESTAMP and JDBC

java.sql.Blob

Clarification for getBytes(int pos, int length)

If the pos (position) argument is greater than the length of the BLOB then an exception is thrown. This matches the semantics of the SQL SUBSTR function.

Clarification for position(byte pattern, int start) and position(Blob pattern, int start)

If the pattern argument has length zero, then the value of start argument will be returned. This matches the semantics of the SQL LOCATE function.

java.sql.Clob

Clarification for getSubString(int pos, int length)

If the pos (position) argument is greater than the length of the CLOB then an exception is thrown. This matches the semantics of the SQL SUBSTR function.

Clarification for position(String searchstr, int start) and position(Clob searchstr, int start)

If the searchstr argument has length zero (the empty string), then the value of start argument will be returned. This matches the semantics of the SQL LOCATE function.

java.sql.Connection

Clarification for createStatement(), prepareStatement() and prepareCall()

CONCUR_UPDATABLE concurrency is supported for TYPE_FORWARD_ONLY and TYPE_SCROLL_INSENSITIVE ResultSet types only.

java.sql.PreparedStatement

Clarification for setAsciiStream(),setBinaryStream(),setCharacterStream()

In Derby, the stream must have the exact amount of data as the length argument, otherwise an exception with the following message 'Input Stream did not have exact amount of data as the requested length' is thrown.
[TUTORIAL3]

Stream parameter values are not re-used. [JDBC3] says in the last paragraph of 13.2.2 that parameters are re-used but nothing special about streams. However javadoc for java.sql.PreparedStatement.clearParamters() says 'In general, parameter values remain in force for repeated use of a statement'. Maybe 'in general' can be interpreted to mean 'except for streams'. Stream parameter values are not re-used and if a stream is re-used, the statement execution will fail with 'Stream has already been read and end-of-file reached and cannot be re-used.'.

java.sql.ResultSet

Extension for getXXX()

Support for data conversion using the getXXX() methods matches Table B-6 of [JDBC3] with this extension:

• If the value column is null, then generally any getXXX method can be used to fetch the null value and a null or a representation of zero will be returned and a subsequent call to wasNull() will return true. It is not recommended that applications take advantage of this liberal conversion of SQL NULL values, use of a getXXX() method that works against NULL and non-NULL values is strongly recommended.

Clarification for deleteRow()

After deleteRow, ResultSet will be positioned right before the next row for TYPE_FORWARD_ONLY and TYPE_SCROLL_INSENSITIVE updatable ResultSets.

Clarification for updateRow()

After updateRow, ResultSet will be positioned right before the next row for TYPE_FORWARD_ONLY updatable ResultSets. The ResultSet remain on the current row for TYPE_SCROLL_INSENSITIVE updatable ResultSets.

Clarification for getAsciiStream()

getAsciiStream() is not recommended to use in Derby, because Derby handles characters as unicode internally , there is no advantage in using this method.

An ASCII character is defined as an eight bit character (range 0x00 to 0xff), see CHAR() function definition by [JDBC3] in appendix C.2.

For character types (Types.CHAR, Types.VARCHAR & Types.LONGVARCHAR), each character in the value is translated to one byte in the ASCII stream:

• Unicode characters in the range 0x0000 to 0x00ff, are translated to a byte with the matching value (0x00 to 0xff).
• Unicode caracters in the range (0x0100 to 0xffff) are translated to 0x3f, ('?', question mark character).

For binary types (Types.BINARY, Types.VARBINARY, and Types.LONGVARBINARY) getAsciiStream() returns a stream with identical contents to that returned by getBinaryStream() on the same column.

Extension for getAsciiStream()

• Types.CLOB columns getAsciiStream() returns a stream with identical contents to that returned by getClob().getAsciiStream() on the same column if the CLOB value is not NULL. If the CLOB value is NULL then null is returned.
• Types.BLOB columns getAsciiStream() returns a stream with identical contents to that returned by getBlob().getBinaryStream() on the same column if the BLOB value is not NULL. If the BLOB value is NULL then null is returned.

Extension for getBinaryStream()

On Types.BLOB columns returns a stream with identical contents to that returned by getBlob().getBinaryStream() on the same column if the BLOB value is not NULL. If the BLOB value is NULL then null is returned.

Clarification for getCharacterStream()

For binary types (Types.BINARY, Types.VARBINARY, and Types.LONGVARBINARY, Types.BLOB) the value is converted to a stream of characters as though it is encoded using UTF-16BE.

Extensions for getCharacterStream()

• On Types.CLOB columns returns a reader with identical contents to that returned by getClob().getCharacterStream() on the same column if the CLOB value is not NULL. If the CLOB value is NULL then null is returned.
• On Types.BLOB columns supported with conversion as a binary type.

Clarification for getString()

For binary types (Types.BINARY, Types.VARBINARY, and Types.LONGVARBINARY, Types.BLOB) getString() returns String containing a two character hexadecimal representation for every byte in a non-null value. The two characters are in the range '0' - '9' and 'a' - 'f'. For NULL values, null is returned.
Note, that this String does not match the contents of the stream returned by getAsciiStream() or getCharacterStream() on the same column.

Extensions for getString()

• On Types.CLOB columns returns a String representing the contents of the CLOB value if it is not NULL. If the CLOB value is NULL then null is returned.
• On Types.BLOB columns supported with conversion as a binary type.

getUnicodeStream()

Not implemented, deprecated by [JDBC3].

Binary column example

If a binary column has been set with the Java byte array containing fours bytes, byte[] data = {0x34, 0x87, 0xc2, 0x1f} then:

• getBinaryStream() returns a java.io.InputStream containing four bytes - 0x34, 0x87, 0xc2, 0x1f
• getAsciiStream() returns a java.io.InputStream containing four bytes - 0x34, 0x87, 0xc2, 0x1f [ same as getBinaryStream() ]
• getCharacterStream() returns a java.io.Reader containing two characters - 0x3487, 0xc21f
• getString() returns a String with eight characters "3487c21f"

Derby's SQL DATE interactions with JDBC

Derby’s SQL DATE type represents a date in the form yyyy-mm-dd with no associated time zone information.

java.sql.Date

A JDBC Date (java.sql.Date) by definition represents a point in time on a given date in a given time zone.

[JDBC3] intends that the point in time for a java.sql.Date object is 00:00 (midnight), but this is not enforced by the class.

JDBC drivers are required to return java.sql.Date objects that are normalized to 00:00 according to the required time zone.

Applications are expected to pass in java.sql.Date instances that are normalized to 00:00 (see section 18.1.1 of [TUTORIAL3]).

Conversion of a JDBC java.sql.Date to a Derby DATE value

1. setDate() without a Calendar object or passing null for a Calendar object:

   • The yyyy-mm-dd values will be calculated from the milli-seconds value of the java.sql.Date instance using a Calendar object set to the time zone of the virtual machine.

   • This yyyy-mm-dd value will match the output of java.sql.Date.toString().

2. setDate() with a Calendar object:

   • The yyyy-mm-dd values will be calculated from the milliseconds value of the java.sql.Date instance using the passed in Calendar.

   • The code for this is
     cal.setTimeInMillis(value.getTime());
     yyyy = cal.get(Calendar.YEAR);
     mm = cal.get(Calendar.MONTH) + 1;
     dd = cal.get(Calendar.DAY_OF_MONTH);

   • This yyyy-mm-dd value may not match the output of java.sql.Date.toString() for the value, since this method always uses the time zone of the virtual machine.

Derby does not require that the application’s java.sql.Date value is normalized to 00:00 according to the required time zone.

In both cases no time zone information is stored with the SQL DATE value.

Conversion of a Derby DATE value to a JDBC java.sql.Date

1. getDate() without a Calendar object or passing null for a Calendar object:

   • A java.sql.Date instance is returned with a millisecond value corresponding to 00:00 on yyyy-mm-dd according to the time zone of the java virtual machine

   • The toString() method of the returned value will return ‘yyyy-mm-dd’

2. getDate() with a Calendar object:

   • A java.sql.Date instance is returned with a millisecond value corresponding to 00:00 on yyyy-mm-dd according to the time zone of the Calendar

   • The toString() method of the returned value may not return ‘yyyy-mm-dd’, since this method always uses the time zone of the virtual machine.

Conversion of a string type to a JDBC java.sql.Date

Three different date formats are built into Derby.

1. (ISO/JIS) yyyy-mm-dd e.g. “1980-03-21”,

2. (IBM USA) mm/dd/yyyy e.g. “03/21/1980”, and

3. (IBM European) dd.mm.yyyy e.g. “21.03.1980”.

If the format of the string matches one of the built in formats then a conversion to a java.sql.Date matches that of a SQL DATE value with value yyyy-mm-dd.

If the string does not match any of the built in formats Derby attempts to use the Java locale specific parser to interpret the string as a date.

Derby's SQL TIME interactions with JDBC

Derby’s SQL TIME type represents a time of day in the form hh:mm:ss with no associated time zone information.

java.sql.Time

A JDBC Time (java.sql.Time) by definition represents a point in time on an unspecified day in a given time zone.

Java.sql.Time extends java.util.date, so it includes a date. [JDBC3] intends that the date stored in a java.sql.Time be Jan 1 1970, but this is not enforced by the class.

JDBC drivers are required to return java.sql.Time objects that are normalized to Jan. 1 1970 according to the required time zone.

Applications are expected to pass in java.sql.Time instances that are normalized to Jan. 1 1970.

Conversion of a JDBC java.sql.Time to a Derby TIME value

1. setTime() without a Calendar object or passing null for a Calendar object

   • The hh:mm:ss values will be calculated from the milli-seconds value of the java.sql.Time instance using a Calendar object set to the time zone of the virtual machine.

   • This hh:mm:ss value will match the output of java.sql.Date.toString().

2. setTime() with a Calendar object

   • The hh:mm:ss values will be calculated from the milliseconds value of the java.sql.Date instance using the passed in Calendar.

   • The code for this is
     cal.setTimeInMillis(value.getTime());
     hh = cal.get(Calendar.HOUR);
     mm = cal.get(Calendar.MINUTE);
     ss = cal.get(Calendar.SECOND);

   • This hh:mm:dd value may not match the output of java.sql.Date.toString() for the value, since this method always uses the time zone of the virtual machine.

Derby does not require that the application’s java.sql.Time value be normalized to Jan 1 1970 according to the required time zone.

In both cases no time zone information is stored with the SQL TIME value.

Conversion of a Derby TIME value to a JDBC java.sql.Time

1. getTime() without a Calendar object or passing null for a Calendar object

   • A java.sql.Time instance is returned with a millisecond value corresponding to hh:mm:ss on Jan. 1 1970 according to the time zone of the java virtual machine

   • The toString() method of the returned value will return ‘hh:mm:ss’.

2. getTime() with a Calendar object

   • A java.sql.Time instance is returned with a millisecond value corresponding to hh:mm:ss on Jan. 1 1970 according to the time zone of the Calendar

   • The toString() method of the returned value may not return ‘hh:mm:ss’, since this method always uses the time zone of the virtual machine.

Conversion of a string type to a JDBC java.sql.Time

Three different time formats are built into Derby:

1. (ISO/EUR) hh.mm.ss e.g. “13.52.03”,

2. (IBM USA) hh:mm [AM|PM] e.g. “1:52 PM”, and

3. (JIS) hh:mm:ss e.g. “13:52:03”.

If the format of the string matches one of the built in formats then a conversion to a java.sql.Time matches that of a SQL TIME value with value hh:mm:ss.

If the string does not match any of the built in formats Derby attempts to use the Java locale specific parser to interpret the string as a date.

Derby's SQL TIMESTAMP interactions with JDBC

Derby’s SQL TIMESTAMP type represents a time of day in the form yyyy-mm-dd hh:mm:ss.fffffffff (nanosecond granularity) with no associated time zone information.

java.sql.Timestamp

A JDBC Timestamp (java.sql.Timestamp) by definition represents a point in time, with nanosecond resolution, in a given time zone.

Conversion of a JDBC java.sql.Timestamp to a Derby TIMESTAMP value

1. setTimestamp() without a Calendar object or passing null for a Calendar object

   • The year, month, day, hour, minute, and second values will be calculated from the milli-seconds value of the java.sql.Timestamp instance using a Calendar object set to the time zone of the virtual machine. The nanosecond value will be calculated from the nanoseconds value of the java.sql.Timestamp.

   • The timestamp component values will match the output of java.sql.Timestamp.toString().

2. setTime() with a Calendar object

   • The year, month, day, hour, minute, and second values will be calculated from the milliseconds value of the java.sql.Date instance using the passed in Calendar. The nanosecond value will be calculated from the nanoseconds value of the java.sql.Timestamp.

   • The code for this is
     cal.setTimeInMillis(value.getTime());
     year = cal.get(Calendar.YEAR);
     month = cal.get(Calendar.MONTH) + 1;
     day = cal.get(Calendar.DAY_OF_MONTH);
     hour = cal.get(Calendar.HOUR);
     minute = cal.get(Calendar.MINUTE);
     second = cal.get(Calendar.SECOND);
     nanos = value.getNanos();

   • This stored timestamp component value may not match the output of java.sql.Timestamp.toString() for the value, since this method always uses the time zone of the virtual machine.

Conversion of a Derby TIMESTAMP value to a JDBC java.sql.Timestamp

1. getTimestamp() without a Calendar object or passing null for a Calendar object

   • A java.sql.Timestamp instance is returned with a nanosecond value corresponding to yyyy-mm-dd hh:mm:ss.fffffffff according to the time zone of the java virtual machine

   • The toString() method of the returned value will return ‘yyyy-mm-dd hh:mm:ss.fffffffff’.

2. getTime() with a Calendar object

   • A java.sql.Time instance is returned with a nanosecond value corresponding to yyyy-mm-dd hh:mm:ss.fffffffff according to the time zone of the Calendar

   • The toString() method of the returned value may not return ‘yyyy-mm-dd hh:mm:ss.fffffffff’, since this method always uses the time zone of the virtual machine.

Conversion of a string type to a JDBC java.sql.Timestamp

Two different timestamp formats are built into Derby:

1. (ISO) yyyy-mm-dd hh:mm:ss[.ffffff]e.g. “1980-10-25 13:01:23.123456”, and

2. (IBM) yyyy-mm-dd-hh.mm.ss[.ffffff]e.g. “1980-10-25-13.01.23.123456”.

Note that only microsecond resolution is supported in converting strings to timestamps.

If the format of the string matches one of the built in formats then a conversion to a java.sql.Timestamp matches that of a SQL TIMESTAMP value with value yyyy-mm-dd hh:mm:ss.ffffff.

If the string does not match any of the built in formats Derby attempts to use the Java locale specific parser to interpret the string as a date.