RE: [xsl] recursive sorting by element name -> Xalan Issue

Subject: RE: [xsl] recursive sorting by element name -> Xalan Issue
From: Wendell Piez <wapiez@xxxxxxxxxxxxxxxx>
Date: Thu, 29 Nov 2007 17:02:01 -0500

At 02:41 PM 11/29/2007, it was written:
No, self:: implies that the context is a node(), and an attribute
doesn't show up on that axis. However, since you're applying templates
to the nodes that will contain the @about attribute, you can omit the
self:: axis altogether. Try this:

<xsl:sort select="+not(@about)"/>

The advice is sound, but the reasoning for it is somewhat garbled.

Attributes do show up on the self:: axis, just as any node does, if the traversal starts there. Try


and you will get back the same set of nodes as from @* without the "self::node()" step.

To disentangle this, understand that:

"@" in XPath is an abbreviation for an axis specifier, "attribute::".

"*" is a wildcard node test in XPath, which retrieves any node of the "primary node type of the axis". The primary node type, in turn, is always an element, except when the axis is the attribute:: axis ('@'), in which case it's an attribute. (Or unless the axis is the namespace axis, which we'll leave aside.) A similar rule is followed for a node test by name, such as "section" -- it returns elements named section, unless the axis is attribute::, in which case we see "attribute::section" or "@section", and get an attribute by that name.

("node()" is a node test that matches any node at all, irrespective of what axis it's on.)

If no axis is given at all, an XPath step defaults to the child:: axis, so "section" with no axis given is the same as "child::section". (This is why select="section" is so different from select="self::section".)

So "self::*" will return an element if the processing context is an element, but nothing if it isn't. Similarly, "self::DictionaryModelDescriptor" returns a DictionaryModelDescriptor element if the processing context is a DictionaryModelDescriptor element, nothing if not.

Similarly, "@*" or "attribute::*" will always traverse the attribute axis from where it starts, and "@about" (aka "attribute::about") will start at the context node, traverse the attribute axis, and return any attribute it finds there with the qualified name "about" (due to XML rules there will never be more than one).

BTW, if you wanted to write for clarity, "+not(@about)" could be expressed "number(not(@about))", or if you preferred you could rely on what Mike reminded us, that false() sorts before true(), and simply say "not(@about)" or "boolean(@about)" depending on which way you need it to sort.

Saying simply "@about" would sort on the value of that attribute, as Scott also says.


Wendell Piez                            mailto:wapiez@xxxxxxxxxxxxxxxx
Mulberry Technologies, Inc.      
17 West Jefferson Street                    Direct Phone: 301/315-9635
Suite 207                                          Phone: 301/315-9631
Rockville, MD  20850                                 Fax: 301/315-8285
  Mulberry Technologies: A Consultancy Specializing in SGML and XML

Current Thread