ah, so it's implemented in the core code?
I'm doing this work on my own wikis, so I can update my own wikis codebase to support this, perhaps.
what version of MW is this implemented on?

barring that approach, I saw some references to obtaining an xml rendering of an entire page. Is that right?
If I could get an xml representation of a rendered page, where each section is noted in the xml somehow, I could then extract the relevent section, with just the additional overhead of pulling the whole page.

However, my efforts to specify the generatexml parameter failed.

Any hints?

Thanks again for your quick response!