[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: Nasty XPath expressions

• To: "'Amy Lewis'" <amyzing@t...>,"'Evan Lenz'" <elenz@x...>
• Subject: RE: Nasty XPath expressions
• From: "Michael Kay" <michael.h.kay@n...>
• Date: Thu, 13 Dec 2001 19:18:03 -0000
• Cc: "'bob mcwhirter'" <bob@w...>,"'Kelvin Yeow'" <whyeow@c...>,<xml-dev@l...>
• Importance: Normal
• In-reply-to: <20011213171714.GB10316@t...>
• Reply-to: <michael.h.kay@n...>

> Umm, no.
>
> Node-sets are ordered sets.  The spec spends (too much?) time
> discussing the ordering of returned sets, which depend upon the axis.
> So it isn't exactly true that they are always in document order
> (particularly if you're using a complex expression that moves in
> multiple axes), but it is true that the nodes are returned in
> the order
> of encounter, as specified by the axis.  Typically, this
> means document
> order.
>
> They are *not* unordered sets.

Wrong. XPath 1.0 node-sets are unordered.

If you read the material in section 2.4 again, you will see that it says an
axis has direction, and that the proximity position of a node in a node-set
depends on the direction of the associated axis. But it never says that
node-sets themselves are ordered, or that the nodes are "returned" in any
particular order. You might find the section easier to understand (I
certainly do) if you think of a node-set derived from a forwards axis as
being a list of nodes in document order, and a node-set derived from a
reverse axis as being a list of nodes in reverse document order, but the
spec very deliberately doesn't describe it that way.

Mike Kay

• References:
  • Re: Nasty XPath expressions
    • From: Amy Lewis <amyzing@t...>

• Prev by Date: ANN: Markup Object Events (MOE)
• Next by Date: Re: XPath node-set order
• Previous by thread: Re: Nasty XPath expressions
• Next by thread: RE: Nasty XPath expressions
• Index(es):
  • Date
  • Thread