Java and XML Basics, Part 3 - Which One is the Better One to Use?
(Page 2 of 9 )
The latter question though is not that easy to answer; while in most cases choosing in between SAX and DOM proves to be a matter of tastes with some programmers, we are going to look at some speed and performance comparisons in between the two of them. While there are complex ways to evaluate the performance of a computer program, we are only going to refer to two main factors:
- Memory consumption
- Execution speed
For this we have modified the SimpleDOMParser3.java and SimpleSAXParser6.java files and added some lines to measure the execution time and the memory usage as it can be seen:
SimpleDOMParser4.java
/**
* Parse the document now
*/
System.gc();
long tmStart = System.currentTimeMillis();
long memStart = Runtime.getRuntime().freeMemory();
try
{
doc = builder.parse( isXML );
}
catch( IOException ioe )
{
System.err.println( "I/O error while reading from file:" );
ioe.printStackTrace();
System.exit( 5 );
}
catch( org.xml.sax.SAXException saxe )
{
System.err.println( "Parsing error:" );
saxe.printStackTrace();
System.exit( 6 );
}
System.gc();
long tmEndParse = System.currentTimeMillis();
long memEnd = Runtime.getRuntime().freeMemory();
System.out.println( "Parsing took : " + (tmEndParse - tmStart) + " msec" );
System.out.println( "Memory occupied : " + (memStart - memEnd) + " bytes" );
SimpleSAXParser7.java
/**
* Parse the document now
*/
System.gc();
long tmStart = System.currentTimeMillis();
long memStart = Runtime.getRuntime().freeMemory();
try
{
sax.parse( isXML, new SimpleSAXParser7() );
}
catch( IOException ioe )
{
System.err.println( "I/O error while reading from file:" );
ioe.printStackTrace();
System.exit( 6 );
}
catch( org.xml.sax.SAXException saxe )
{
System.err.println( "Parsing error:" );
saxe.printStackTrace();
System.exit( 7 );
}
System.out.println();
System.gc();
long tmEnd = System.currentTimeMillis();
long memEnd = Runtime.getRuntime().freeMemory();
System.out.println( "Parsing took " + (tmEnd - tmStart) + " msec" );
System.out.println( "Memory occupied " + (memStart - memEnd) + " bytes" );
As you can see we don't do any fancy time estimation, just base everything on the computer clock and measure the number of milliseconds it takes for the process to finish -- of course in this process we don't just execute parsing code but a few other things as well (check for exceptions etc.), but we consider the time spent on executing these instructions as irrelevant -- and even if it were, nearly the same code appears in both files!
Also, to have an estimate of the memory occupied, we just compare the available free memory (in bytes) before and after the parsing. Again, in this process memory can be occupied not just by the variables and structures involved in parsing but by Java internal structures too (buffers, stacks, unclaimed memory blocks etc.). To make the result look closer to reality, we are suggesting to the interpreter that the garbage collection should be kicked off before and right after the parsing -- this will give a more accurate figure of the memory available. Of course, there is no way to enforce garbaged memory recollection, however, as we are running in a single-user single-threaded non-stressful environment, we can assume that in most cases garbage collection will take place when we have queued up the request.
Next: Running the Parser >>
More XML Articles
More By Liviu Tudor
