Roman Rokytskyy wrote:

>It's been a while since we have published a document describing possible
>architecture to support external procedures written in different languages.
>Today we have released first preview version of Java support in Firebird.
>
>
>

It would be both easier to use, easier to implement, and easier to
maintain if you actually passed the parameters necessary for triggers,
udfs, and procedures rather than requiring them to call static procedure
to pick these things out of the ether. A trigger, for examples, needs a
connection, a pre-update records (possibly null), a post-update record,
and a trigger operation type. These should be passed directly to the
trigger.

In Netfrastruture, which has had an embedded Java implementation for
five years, the signature for a trigger method is:

public static void methodName (Connection connection, Record
oldVersion, Record newVersion, int type)

where Record:

public interface Record
{
public static final int PreInsert = 1;
public static final int PostInsert = 2;
public static final int PreUpdate = 4;
public static final int PostUpdate = 8;
public static final int PreDelete = 16;
public static final int PostDelete = 32;
public static final int PreCommit = 64;
public static final int PostCommit = 128;

public String getSchemaName() throws SQLException;
public String getTableName() throws SQLException;

public int getColumnId (String columnName) throws
SQLException;
public int nextColumnId (int columnId) throws SQLException;
public int nextPrimaryKeyColumn (int columnId) throws
SQLException;
public boolean isChanged (int columnId, Record record) throws
SQLException;
public boolean wasNull () throws SQLException;

public String getColumnName (int columnId) throws SQLException;
public int getColumnType (int columnId) throws SQLException;
public int getPrecision (int columnId) throws SQLException;
public int getScale (int columnId) throws SQLException;

public String getString(String columnName) throws SQLException;
public boolean getBoolean(String columnName) throws SQLException;
public byte getByte(String columnName) throws SQLException;
public short getShort(String columnName) throws SQLException;
public int getInt(String columnName) throws SQLException;
public long getLong(String columnName) throws SQLException;
public float getFloat(String columnName) throws SQLException;
public double getDouble(String columnName) throws SQLException;
public byte[] getBytes(String columnName) throws SQLException;
public java.sql.Timestamp getTimestamp(String columnName) throws
SQLException;
public java.sql.Date getDate(String columnName) throws SQLException;
public java.sql.Time getTime(String columnName) throws SQLException;
public netfrastructure.sql.NfsBlob getBlob (String columnName)
throws SQLException;
public netfrastructure.sql.NfsBlob getClob (String columnName)
throws SQLException;

public void setNull(String columnName, int sqlType)
throws SQLException;
public void setString(String columnName, String value)
throws SQLException;
public void setBoolean(String columnName, boolean value)
throws SQLException;
public void setByte(String columnName, byte value) throws
SQLException;
public void setShort(String columnName, short value)
throws SQLException;
public void setInt(String columnName, int value) throws
SQLException;
public void setLong(String columnName, long value) throws
SQLException;
public void setFloat(String columnName, float value)
throws SQLException;
public void setDouble(String columnName, double value)
throws SQLException;
public void setBytes(String columnName, byte[] value)
throws SQLException;
public void setTimestamp(String columnName,
java.sql.Timestamp value) throws SQLException;
public void setDate(String columnName, java.sql.Date
value) throws SQLException;
public void setTime(String columnName, java.sql.Time
value) throws SQLException;
public void setBlob(String columnName, java.sql.Blob
value) throws SQLException;
public void setClob(String columnName, java.sql.Clob
value) throws SQLException;

public String getString(int columnId) throws SQLException;
public boolean getBoolean(int columnId) throws SQLException;
public byte getByte(int columnId) throws SQLException;
public short getShort(int columnId) throws SQLException;
public int getInt(int columnId) throws SQLException;
public long getLong(int columnId) throws SQLException;
public float getFloat(int columnId) throws SQLException;
public double getDouble(int columnId) throws SQLException;
public byte[] getBytes(int columnIndex) throws SQLException;
public java.sql.Timestamp getTimestamp(int columnId) throws
SQLException;
public java.sql.Date getDate(int columnId) throws SQLException;
public java.sql.Time getTime(int columnId) throws SQLException;
public netfrastructure.sql.NfsBlob getBlob(int columnId) throws
SQLException;
public netfrastructure.sql.NfsClob getClob(int columnId) throws
SQLException;

public void setNull(int columnId, int sqlType) throws
SQLException;
public void setString(int columnId, String value) throws
SQLException;
public void setBoolean(int columnId, boolean value)
throws SQLException;
public void setByte(int columnId, byte value) throws
SQLException;
public void setShort(int columnId, short value) throws
SQLException;
public void setInt(int columnId, int value) throws
SQLException;
public void setLong(int columnId, long value) throws
SQLException;
public void setFloat(int columnId, float value) throws
SQLException;
public void setDouble(int columnId, double value) throws
SQLException;
public void setBytes(int columnId, byte[] value) throws
SQLException;
public void setTimestamp(int columnId, java.sql.Timestamp
value) throws SQLException;
public void setDate(int columnId, java.sql.Date value)
throws SQLException;
public void setTime(int columnId, java.sql.Time value)
throws SQLException;
public void setBlob(int columnId, java.sql.Blob value)
throws SQLException;
public void setClob(int columnId, java.sql.Clob value)
throws SQLException;

}

A key issue for triggers is metadata access. Existing Firebird triggers
are necessarily single table, so built in knowledge of metadata isn't a
problem. Java triggers, on the other hand, are naturally table
independent, so building in a mechanism for rapid, context sensitve
access to metadata is critical.

The methods "nextColumnId" and "nextPrimaryKeyColumn" allow rapid
iteration through all fields or primary keys, for example:

for (int n = -1; (n = record.nextPrimaryKeyColumn (n)) >= 0;)
key += record.getString (n) + "^";

The rest of the methods are necessary for metadata and data access and
data update.



--

Jim Starkey
Netfrastructure, Inc.
978 526-1376