There is a known problem with new XSLT processor in just released Microsoft .NET 2.0 - XslCompiledTransform class doesn't support pull mode XSLT - outputting XSLT result in XmlReader form as obsolete XslTransform class did. I'm happy to announce that the problem is solved now. Sergey Dubinets from the Microsoft XML Team has developed XslReader implementation which allows to read XSLT output via XmlReader in a very efficient way - i.e. without full output buffering. It's concurrent writer-reader bridge, which runs XSLT in another thread and lets it to produce another output portion whenever reader wants to read more.
And Sergey kindly donated his code to the Mvp.Xml project, so it will be part of the Common module of Mvp.Xml library v2.0, which is going to be released really really really really soon. Kudos to Sergey! Stay tuned.
Sorry, it's not available yet, but it's a matter of days now.
Thanks for the heads up!
Is this code available anywhere yet? As you can see from the comments on the bug report it's something I definitely need. I'm very happy that someone has stepped up to "do the right thing"... (and that the Mvp.Xml project exists as a ship vehicle)