Glen, I'd like this change reverted. Here's the justification:
VM warm-up (3 runs) VM warm-up finished. unbuffered: 766ms unbuffered: 781ms unbuffered: 766ms unbuffered: 750ms unbuffered: 703ms unbuffered: 719ms Average (unbuffered): 747ms buffered: 609ms buffered: 594ms buffered: 562ms buffered: 594ms buffered: 578ms buffered: 594ms Average (buffered): 588ms Difference: 22% Not nothing. The test is run on a 110KB FO file with some SVG graphics in it under FOP HEAD. Buffering the OutputStream should be a best practice if the user is not writing the output to a ByteArrayOutputStream. See also: http://xml.apache.org/fop/embedding.html#performance Thanks! On 15.07.2004 00:45:27 gmazza wrote: > gmazza 2004/07/14 15:45:27 > > Modified: examples/embedding/java/embedding ExampleFO2PDF.java > Log: > Removed BufferedOutputStream line as unneeded in this case (Unsure if helpful > for performance.) > > Revision Changes Path > 1.10 +0 -1 xml-fop/examples/embedding/java/embedding/ExampleFO2PDF.java > > Index: ExampleFO2PDF.java > =================================================================== > RCS file: /home/cvs/xml-fop/examples/embedding/java/embedding/ExampleFO2PDF.java,v > retrieving revision 1.9 > retrieving revision 1.10 > diff -u -r1.9 -r1.10 > --- ExampleFO2PDF.java 14 Jul 2004 22:42:29 -0000 1.9 > +++ ExampleFO2PDF.java 14 Jul 2004 22:45:27 -0000 1.10 > @@ -59,7 +59,6 @@ > > // Setup output > out = new java.io.FileOutputStream(pdf); > - out = new java.io.BufferedOutputStream(out); > driver.setOutputStream(out); > > // Setup JAXP using identity transformer Jeremias Maerki
