XSLT is also a rather
arcane technology.
I am trying to make up my mind whether it is a really cool idea or a really sucky one. Though I think I understand the underlying motivations, every time I try to actually use the thing I cannot help but feeling that having a transformation API with appropriate language bindings (rather than a document specification that is also a programming language that requires its own processor...) would be way better.
Yesterday, as I was trying to put msxml's XSLT processor (the Microsoft implementation of the standard) back into working order, a coworker asked me:
"Do you think we'll be able to get this stuff to work before XML becomes outdated?"
A statement which, at that point, exactly reflected my feelings.
A separate point of concern lies with the CS lingo that has been generously sprinkled on what is basically a trivial tree navigation/selection: the parent axis, the attribute axis... this kind of verbiage puts the entire XSLT shebang at serious death by jargon risk. It looks fatigued already...