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

Re: XPath for Infoset extensions [was Annotations in

• To: John Cowan <jcowan@r...>
• Subject: Re: XPath for Infoset extensions [was Annotations in
• From: Jeni Tennison <jeni@j...>
• Date: Mon, 7 Oct 2002 15:58:06 +0100
• Cc: wperry@f... (W. E. Perry), xml-dev@l... (XML DEV)
• In-reply-to: <200210071440.KAA15423@m...>
• Organization: Jeni Tennison Consulting Ltd
• References: <200210071440.KAA15423@m...>
• Reply-to: Jeni Tennison <jeni@j...>

Hi John,

> There's a confusion of levels lurking here. An infoset may be
> represented in two different ways: as the native XML from which it
> was derived, or with some other representation particular to
> infosets. This latter may or may not be an application of XML.

Thank you for articulating it like that! This is what I've been trying
to say about LMNL -- you can represent a LMNL data model / infoset in
a "native" form, as LMNL syntax, or with some other representation
peculiar to the LMNL data model. The latter may or may not be an
application of LMNL. We have, though, defined a standard "reified LMNL
layer", which is an application of LMNL that represents a LMNL data
model (and a few other things that are useful in LMNL syntax).

> The first representation has the advantage that it is simple and
> clear; it has the disadvantage that it cannot in general represent
> infosets that contain non-standard item types or properties. The
> second representation is generally quite opaque, but has the
> advantage that it can easily be extended to represent enhanced
> infosets.
>
> SLAIX is a general-purpose representation of infosets that has the
> advantages of uniformity, extreme extensibility, and being an
> application of XML. Its disadvantages are that it is verbose,
> grotesque, and unreadable. See
> http://lists.xml.org/archives/xml-dev/200208/msg00875.html

Having a short, elegant and readable representation for moving around
augmented Infosets was one my primary reasons for pursuing LMNL.

In LMNL syntax, the Infoset represented by the XML document:

  <book>
    <author sort="brownjohn">John Brown</author>
    <title>My Life</title>
  </book>

would look like:

  [book}
    [author [sort}brownjohn{]}John Brown{author]
    [title}My Life{title]
  {book]

If you wanted to augument this infoset, for example to say that the
xsi:type of the [sort] annotation was xs:token, you could do that by
adding an annotation to the [sort] annotation:

  [!ns xsi="http://www.w3.org/2001/XMLSchema-instance"]
  [book}
    [author [sort [xsi:type}xs:token{]}brownjohn{]}John Brown{author]
    [title}My Life{title]
  {book]

Cheers,

Jeni

---
Jeni Tennison
http://www.jenitennison.com/

• References:
  • Re: XPath for Infoset extensions [was Annotations in
    • From: John Cowan <jcowan@r...>

• Prev by Date: Re: heritage (was Re: SGML on the Web)
• Next by Date: Re: heritage (was Re: SGML on the Web)
• Previous by thread: Re: XPath for Infoset extensions [was Annotations in
• Next by thread: Re: XPath for Infoset extensions [was Annotations in XPath-NG?]
• Index(es):
  • Date
  • Thread