- From: Joseph Kesselman <keshlam@us.ibm.com>
- Date: Fri, 18 Oct 2002 10:04:07 -0400 (EDT)
- To: public-qt-comments@w3.org
One large source of confusion over the XPath spec is that the data model's pseudo-code syntax, due to its use of prefixed function names, looks like it's something that ought to be callable from an XPath expression. In particular, if you don't take time to read and understand Section 2, it looks like the constructors ought to be invokable from within an XPath expression. This has caused at least two quite sophisticated W3C participants that I know of to believe that XPath can produce new document fragments all by itself, rather than simply providing a base for XQuery/XSLT. I really can't blame them for the misunderstanding. The similarity between the syntax of the pseduocode and XPath itself really is *extremely* confusing; I misread it myself earlier today. STRONG SUGGESTION: At the very least, replace the "dm:" prefix on the pseudocode with "DM-" -- or, better yet "PSEUDO-", to make very clear that these operations are *not* intended/promised to be directly invocable, or even necessarily directly coded at the implementation level. Yes, I know, section 2 states explicitly that this is all pseudocode descriptions of behavior. But when flipping between the Data Model/Functions and Operators/XPath/XSLT/XQuery documents, it is simply too easy to get confused about where you are and thus whether a given name is invocable syntax or pseudocode semantics. (Ideally, I would suggest that -- given this problem of context -- a less code-like pseudocode be adopted, so the distinction is obvious at a glance. But changing the names to add a PSEUDO- flag would help tremendously!) ______________________________________ Joe Kesselman / IBM Research ______________________________________ Joe Kesselman / IBM Research
Received on Friday, 18 October 2002 12:32:30 UTC