An important feature of the Logging API is
that the logging methods called by applications never throw
exceptions: it is not reasonable to expect programmers to nest all
their logging calls within try/catch blocks, and
even if they did, there is no useful way for an application to
recover from an exception in the logging subsystem. Since handler
classes such as FileHandler are inherently subject
to I/O exceptions, the ErrorManager provides a way
for a handler to report an exception instead of simply discarding it.
All Handler objects have an instance of
ErrorManager associated with them. If an exception
occurs in the handler, it passes the exception, along with a message
and one of the error code constants defined by
ErrorManager to the error( )
method. error( ) writes a message describing the
exception to System.err, but does so only the
first time it is called: the expectation is that a
Handler that throws an exception once will
continue to throw the same exception with each subsequent log
message, and it is not useful to flood System.err
with repeated error messages. You can of course define subclasses of
ErrorManager that override error(
) to provide some other reporting mechanism. If you do
this, register an instance of your custom
ErrorManager by calling the
setErrorManager( ) method of your
Handler.
public class ErrorManager {
// Public Constructors
public ErrorManager( );
// Public Constants
public static final int CLOSE_FAILURE; =3
public static final int FLUSH_FAILURE; =2
public static final int FORMAT_FAILURE; =5
public static final int GENERIC_FAILURE; =0
public static final int OPEN_FAILURE; =4
public static final int WRITE_FAILURE; =1
// Public Instance Methods
public void error(String msg, Exception ex, int code); synchronized
}
