btw I'm remembering when Jim Bolla has done that work. There was a big perf. improv. using serialized classes, even using validation, compared with a "normal" XML parser without serialization (see NH1.2.x). I'm working to remove the XML parser and use only deserialized classes; in the NH2 series, we having a nice mix-salad. I'm in the middle of some personal stuff but, hopefully, I can work on it before the end of the year (occidental year).
Thanks Richard. 2009/11/29 Richard Brown (gmail) <[email protected]> > Hi, > > I sent this earlier in the week, but it didn't make it to the dev group for > some reason ... trying again. > > (It was in my g-mail sent items ok?) > > *From:* Richard Brown (gmail) <[email protected]> > *Sent:* Sunday, November 22, 2009 10:36 AM > *To:* [email protected] > *Subject:* Re: [nhibernate-development] Info about mappings XSD-validation > > Hi Fabio, > > Confirmed, and I agree bug-finding would be much harder without it. > > I tried removing the validating reader from the > Configuration.LoadMappingDocument(XmlReader, string) method, and just read > the document straight from the hbmReader passed in. > > Assuming I've modified the correct code, I got no observable difference > when running the tests 3 times from "nant test" (notwithstanding 2 tests > fail when the validation is removed): > > Without validation: 232.4s, 239.1s, 228.3s > With validation: 228.1s, 255.2s, 230.5s > > The results I get for running the suite vary wildly (probably my crappy > laptop). I suspect we'd need a different strategy if we want to test > performance. (Perhaps a dedicated fixture/suite for performance, and/or use > of profiling API or logging to determine bottlenecks.) > > Cheers, > Richard > > > *From:* Fabio Maulo <[email protected]> > *Sent:* Sunday, November 22, 2009 6:28 AM > *To:* [email protected] > *Subject:* [nhibernate-development] Info about mappings XSD-validation > > I have tried to disable the schema validation loading mappings. > Can you confirm the non-existence of performance improvement ? > > > -- Fabio Maulo
