Package mediautil.gen.directio

This is a generic package which helps in reducing Disk IO.

See:
Description

Interface Summary

ByteCounter	This Interface provides for tracking the number of bytes read/written.
IterativeReader	Interface specifying read from a stream in small chunks.
IterativeWriter	Interface specifying write to a stream in small chunks.

Class Summary

InStreamFromIterativeWriter	This class enables reading off directly from an IterativeWriter.
OutStreamToIterativeReader	This class enables writing directly to an IterativeReader which will read the data written as an InputStream.
SplitInputStream	This class enables the Sharing of an InputStream (Say from a file) by One or more IterativeReaders with a main reader.

Package mediautil.gen.directio Description

This is a generic package which helps in reducing Disk IO. The features provided by this package are similiar to a pipe within the same thread.

LLJTran supports the features provided by this package.

The class supporting directio features has to support the IterativeReader and/or IterativeWriter interfaces. So it has to be able to read and/or write in small chunks as specified by the nextRead(..)/nextWrite(..) methods in these Interfaces. This involves some code reorganization See the LLJTran documentation and source for example. How this was done in LLJTran is as follows:

• The read/write method is split between an init method and a nextRead/nextWrite method. The actual read/write method calls the init method and then repeatedly calls the nextRead/nextWrite method with large number of bytes till it returns IterativeReader.STOP.
• All local variables in the original read/write method are transferred to a class to make them persistent across calls. For the nextRead/nextWrite method different stages of Reading or Writing are converted similiar to a State Machine. Loops are converted so that you can re-enter the loop and also break from a loop on completion of number of bytes. The init method has common initializations.
• Any read or write call which possibly reads/writes a big array (Like Appx in LLJTran) is split into iterative calls.

The following can be used with Classes implementing IterativeReader:

• SplitInputStream for sharing an Input source from a file with another Object reading from the file.
• OutStreamToIterativeReader enabling an Object to write directly to an IterativeReader so that the output need not be written to a temporary file and then be read.

The following can be used with Classes implementing IterativeWriter:

• InStreamFromIterativeWriter enabling an Object to read directly from an IterativeWriter so that the output need not be written to a temporary file and then be read.