W3C home > Mailing lists > Public > www-xml-infoset-comments@w3.org > October to December 2000

Comments on XML Infoset 20 December 2000 WD

From: Steve Rowe <sarowe@textwise.com>
Date: Fri, 29 Dec 2000 15:40:14 -0500
To: <www-xml-infoset-comments@w3.org>
Message-ID: <NCBBJHNPJKLDEEEJDICPKEPCCMAA.sarowe@textwise.com>
The nature of stringness, or, The synonymy of nullity and emptiness?

I second Lars Marius Garshol's motion[1]:

   Defining and coordinating the use of the terms
   'null' and 'not present' would probably be useful.

I see that 'not present' is not present in the current draft, but the
term 'empty string' should be added to his list (used e.g. in Sections
2.2 and 2.3, item 3 [prefix]; and in Section 2.4, item 2 [content]).

Also, given that most properties appear to be of type "string", there
should be some description of the correspondence, if any (!) between
strings and character infoitems.  (See "Inconsistency ..." below.)

Pre-validation infoset state

In Section 2.3 Attribute Information Items, item 6 [attribute type],
the legitimate values exclude the pre-validation case, where null
would be appropriate.

In Section 2.6 Character Information Items, item 2 [element content
whitespace], a pre-valid infoset cannot determine whether whitespace
appears in element content.  This flag cannot therefore be boolean, as
implied by the specification.  All three possible states (yes, no, and
unknown) should be represented here.

Inconsistency between (2.3 and 2.15) and (2.6, 2.13 and 2.14)

To amplify and multiply addificate to what Martin Gudgin says[2]:

   Section 2.3 - Attribute Information Items states
   that attributes do not have a [children] property
   and gives some justification for this decision

   Section 2.6 - Character Information Items. In the
   description of the parent property refers to the
   [children] property of 'element, attribute and
   namespace attribute' information items.

In addition, Namespace Information Items (Section 2.15) do not have a
[children] property, while Sections 2.13 and 2.14 (Entity Start and
End Marker Information Items, respectively) refer to the [children] of
AIIs and NSIIs.

Political correctitudinality, gender-wise

In the Appendix A, Informative References, the DOM level 1 editors not
mentioned by name are referred to by means of the Latin phrase "et
alii."  If all of the editors were male, this would be correct, "alii"
being the masculine plural form.  However, such is not the case;
Lauren Wood, the DOM WG chair and one of the DOM editors, is a woman.
Thus the correct form is the neutral plural "et alia" (or simply "et
al.," which neatly avoids the issue entirely.)

CDATA sections

It should be noted that removing markers for CDATA sections would
degrade the round-trippability of an infoset, i.e.
MakeInfoset(Serialize(infoset)) != infoset, because of the predefined
entity references which serialization could introduce.  In the case
where the predefined entities are redeclared with different
replacement texts, CDATA sections might be the only way to represent
the predeclared entities' texts (if no replacements for the
predeclared entities are declared and the re-serialization replicates
the declared entities).

Finally, some typos

2.1:1:Sentence[3]:s/(comment information item) each/$1 for each/
2.1:8:s/An string/A string/
2.2:Sentence[1]:s/(There is) a element/$1 an element/
2.2:5:Sentence[2]:s/element, If/element. If/
2.3:Sentence[1]:s/(There is) a attribute/$1 an attribute/
2.5:2:s/unexpand entity/unexpanded entity/
3:Paragraph[2]:s/Specification/A specification|Specifications/

Received on Friday, 29 December 2000 15:41:05 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 23:08:00 UTC