|CONTENTS | PREV | NEXT|
The Image I/O API is designed as a pluggable framework into which any developer may add their own "plug-ins." A plug-in is defined as a set of Java language classes which may be loaded into the API at run-time and which add functionality to the AP I. In the context of the Java Image I/O API, a plug-in may provide the ability to read image data from a new file format, to write image data in a new format, to "transcode" non-image metadata between two formats, or to read or write streaming dat a from or to a new data source or sink. A plug-in may also provide support for the same format as another plug-in, perhaps providing better performance, more features, or a different view of the data stored by the format.
In order for plug-ins to be added to a running JavaTM Virtual Machine*, they must be compiled into Java bytecode files (class files). These class files will contain the code for subclasses of various classes defined by the API. For example, a plug-in that pr ovides the capability to read images will include a new subclass of the abstract
javax.imageio.ImageReaderclass defined by the API. The usual reversed Internet domain name convention may be used to guarantee uniqueness of class names.
Because loading and instantiating an entire plug-in may be expensive, an additional class is used as a "stand-in" to provide information about the plug-in. This class may be used, for example, to determine what formats are handled by the plug-in p rior to actually instantiating the plug-in.
The "stand-in" object is lightweight enough that it can be loaded and a single instance instantiated every time the API is used within a given invocation of the Java virtual machine. This pattern, in which a small class is used to provide informat ion about an available service, is referred to as a "service provider interface."
Copyright © 1993, 2011, Oracle and/or its affiliates. All rights reserved.