Hi Georg On 18.02.2009 10:13:21 Georg Datterl wrote: > Hi Jeremias, > > > I've also made performance measurements as part of this effort which > > highlights why it was done in the first place: > > http://people.apache.org/~jeremias/fop/benchmark-2009-02-13/ > > I'm probably missing something important here, but regarding the first graph: > > Isn't "direct" what ordinary users of fop do? Take a fo-file and then render > it?
Yes. > Isn't "direct-via-if" what ordinary users of fop will do in the future, if IF > is the default? Yes. > Isn't the work done in "direct-via-if" the sum of the work done in "to-if" > and "from-if"? No. "to-if" renders the FO and uses IFSerializer (called by IFRenderer) to write an intermediate file. "from-if" parses the intermediate file (using IFParser) and generates a series of calls against an IFDocumentHandler and IFPainter implementation. So: "direct-via-if" = "to-if" + "from-if" - serializing IF - parsing IF or "to-if" + "from-if" = "direct-via-if" + serializing IF + parsing IF The main motivation for the new IF is the fact that "serializing AT XML" and especially "parsing AT XML" is very costly. See also http://people.apache.org/~jeremias/fop/renderer-design-new.png which shows the two different paths. > And, if all the above answers are yes, why is the performance gain > noticed in "from-if" not more obviously related to the performance gain > in "direct-via-if"? There was a "no" above so this is not applicable anymore. I hope I could clear up the gap. Jeremias Maerki