OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: We need an XPath API

At 15:56 2001 03 06 -0500, Charles Reitzel wrote:
>Question: I just read the XPath spec (http://www.w3.org/TR/xpath.html) and
>checked the errata.  Where does it say that CDATA sections are supressed.
>The Infoset (current and previous versions) will include CDATA start/end
>nodes as element children.  My reading of the XPath data model is that it
>just uses the Infoset's notion of element children.

[Ordinarily, I delete all individuals in the recipient list and
leave just the xml-dev adddress, but for reasons I cannot fathom,
this message has such a mess of addresses that I cannot figure out 
if everyone is on xml-dev or not, so I'm just leaving them alone.
But if anyone replies to this message, please at least delete my
address--I am on xml-dev, and one copy is plenty for me.]

Some points of information:

Earlier drafts of the Infoset spec did have cdata section start/end
markers.  The Last Call announcement that I made on xml-dev over 
a month ago [1] pointed out the issue of cdata section markers
as a Specific feedback request issue with pointers to both pro
and con positions.  During Last Call, we got a preponderance of
opinion for omitting cdata section start/end markers from the
infoset, and that is the decision that the XML Core WG took at
our comment resolution discussion last week.  So the Infoset
spec will not include cdata section start/end markers.

The XPath data model [2] does not include cdata section markers.
The spec quite clearly lists 7 types of nodes in the data model:
root, element, text, attribute, namespace, processing instruction,
comment.  So cdata sections are not in the model.  Later, under
Text Nodes [3], it discusses how to convert cdata sections into
regular text nodes when mapping to the XPath data model.

The XPath 1.0 data model does not use the Infoset (mostly because
the Infoset wasn't ready in time), though efforts have been made
to keep the models generally consistent with one another.  But the
current definition of the Infoset as a "set of definitions for use 
in other specifications" means that the omission of some piece of 
information from the infoset doesn't prevent some other spec from 
making use of that information.


[1] http://lists.xml.org/archives/xml-dev/200102/msg00064.html
[2] http://www.w3.org/TR/xpath#data-model
[3] http://www.w3.org/TR/xpath#section-Text-Nodes