This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.

Bug 6553 - XML 1.0 and 1.1 references are inconsisten
Summary: XML 1.0 and 1.1 references are inconsisten
Status: CLOSED FIXED
Alias: None
Product: XML Schema
Classification: Unclassified
Component: Datatypes: XSD Part 2 (show other bugs)
Version: 1.1 only
Hardware: PC Windows XP
: P2 normal
Target Milestone: CR
Assignee: David Ezell
QA Contact: XML Schema comments list
URL: http://www.w3.org/TR/2009/WD-xmlschem...
Whiteboard:
Keywords: decided
Depends on:
Blocks:
 
Reported: 2009-02-09 16:50 UTC by Henry S. Thompson
Modified: 2009-04-22 09:19 UTC (History)
2 users (show)

See Also:


Attachments

Description Henry S. Thompson 2009-02-09 16:50:55 UTC
Our references to XML 1.1 (in both parts 1 and 2) are to the undated version.  Our references to XML 1.0 (in both parts 1 and 2) are to a dated version, namely the 3rd edition, which is now not 1 but 2 editions behind.

I think we should be pointing to non-dated versions throughout, and making clear in sections 1.4 and 1.3 respectively that it is implementation-defined what edition(s) a particular implementation is based on.

I'm prepared to argue that this is an editorial change, but I'd particularly like to hear from Sandy and Mike on this matter (i.e., with respect to what edition of XML their parsers implement).
Comment 1 Michael Kay 2009-02-09 17:01:47 UTC
>(i.e., with respect to what edition of XML their parsers implement).

Real life is more complicated! The Saxon schema validator isn't tied to any particular XML parser (any SAX or StAX parser can be used, or on the .NET platform, the Microsoft parser). Stuff that's defined in XML, like element and attribute names, depends on the user's choice of parser. Stuff that's defined in XML Schema, like the legitimate contents of an xs:NCName or the meaning of \i and \c in regular expressions, depends on a switch to the schema processor which is named xmlversion:1.0|1.1 and which looks as if it might have to accept a third value xmlversion:1.0.5. I have no way of forcing the setting at the parser level to be consistent with the setting at the schema processor level.

Michael Kay
Comment 2 Henry S. Thompson 2009-02-09 17:33:05 UTC
So, wrt the bit where

  xmlversion:1.0|1.1

matters, what edition do you refer to?  The version of Unicode referenced changed between XML 3rd edition and XML 4th edition, for instance, which in turn would in principle affect some datatypes, right?
Comment 3 Michael Kay 2009-02-09 17:42:16 UTC
>The version of Unicode ...

I haven't got that under control at the moment. Some places where Saxon has its own Unicode data it's based on 4.0, some places it relies on Java and that depends which version of the JVM the user's running. It's never been an issue with users. The XML version has never been issue with users either, but it is an issue with conformance tests.
Comment 4 David Ezell 2009-02-13 17:25:13 UTC
 RESOLUTION: citations to W3C specs should give date of publication (good bibliographic practice), cite undated URI, and add a note saying "The edition cited is the one current at the date of publication of this specification.  Implementations may conforms to the edition cited or any later edition(s); it is implementation-defined which."
Comment 5 Henry S. Thompson 2009-02-20 16:36:27 UTC
Typo above, resolution from the minutes reads:

 "RESOLUTION: citations to W3C specs
    should give date of publication (good bibliographic practice),
    cite undated URI, and add a note saying "The edition cited is
    the one current at the date of publication of this
    specification. Implementations may follow the edition cited or
    any later edition(s); it is implementation-defined which."
Comment 6 C. M. Sperberg-McQueen 2009-04-13 00:34:12 UTC
A pair of wording proposals intended to resolve this issue are at

  http://www.w3.org/XML/Group/2004/06/xmlschema-1/structures.b6553.html
  http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b6553.html
  (member-only links)

In order to keep these proposals simple and easy to evaluate, these 
proposals do not attempt to address the various shortcomings and weaknesses
of our references (inconsistent bibliographic practice, misplacement of 
normative references into non-normative list and vice versa).  Those
weaknesses will either be addressed in a separate proposal or not at all.

Comment 7 Sandy Gao 2009-04-13 15:35:18 UTC
Comments to proposals cited in comment #6:

Structures:

1. References to "XML Schema 2nd Edition" and "XSLT 2.0"

There are deletions for the old dated link, but the undated link isn't marked as addition. Diff markup issue?

2. Reference to "XML Schema: Datatypes"

It's shown as "../datatypes/..." I *think* it'll show correctly when public drafts are prepared, but want to make sure that's the case.

Datatypes:

1. Reference to "XML"

There are 2 commas after "(Second Edition)" (not introduced by this proposal)

2. Reference to "XML Schema: Structures"

(Similar to above) Shown as ../structures/...

3. References to "XPath 2.0" and "XSL"

There are deletions for the old dated link, but the undated link isn't marked as addition.
Comment 8 John Arwe 2009-04-13 15:47:37 UTC
(In reply to comment #6)
>   http://www.w3.org/XML/Group/2004/06/xmlschema-1/structures.b6553.html
The xptr ref now has an incorrect url (to xpath)

>   http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b6553.html
XML Schema Reqts appears to have no content.

Comment 9 C. M. Sperberg-McQueen 2009-04-13 16:26:30 UTC
Responding to comment #7 and comment #8:

Sandy Gao writes in comment #7:
> Structures:

> 1. References to "XML Schema 2nd Edition" and "XSLT 2.0"

> There are deletions for the old dated link, but the undated link
> isn't marked as addition. Diff markup issue?

Yes, thanks; fixed in the source now.

> 2. Reference to "XML Schema: Datatypes"

> It's shown as "../datatypes/..." I *think* it'll show correctly when
> public drafts are prepared, but want to make sure that's the case.

I'd like to, too.  But yes, that's the intent.  

I'm pretty sure it works correctly.  The URI comes from entities
defined at document production time, and attempts to provide the URI
for the version of Datatypes which corresponds most closely to the
version of Structures it appears in.  And this reference has been
handled this way since the first public working draft of XSD 1.1,
without any comments.  (That is NOT proof that there has never been a
problem.  But I hope it's suggestive.)

> Datatypes:

> 1. Reference to "XML"

> There are 2 commas after "(Second Edition)" (not introduced by this
> proposal)

There are a number of problems in the references.  After some
consideration and sould searching, I split the proposal into two
parts.  The first part, to resolve issue 6553, does just what 6553
asks, and no more.  

All other fixes -- and they are numerous -- have been postponed to a
separate editorial proposal which I hope to get to this week.

> 2. Reference to "XML Schema: Structures"

> (Similar to above) Shown as ../structures/...

Similar logic to the sister reference from Structures to Datatypes.  I
thought about changing it to a static reference the the TR page copy,
but decided against it, in order to have a cleaner date and to avoid
forgetting to update in the press of publication.

> 3. References to "XPath 2.0" and "XSL"

> There are deletions for the old dated link, but the undated link
> isn't marked as addition.

Thank you.  Error in the diff markup.  By the time I reached
these references I was very bored.


Comment #8

> (In reply to comment #6)
> http://www.w3.org/XML/Group/2004/06/xmlschema-1/structures.b6553.html
> The xptr ref now has an incorrect url (to xpath)

Thank you; I've fixed it in the source.

>  http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b6553.html
>  XML Schema Reqts appears to have no content.

Yes.  As explained above, I decided to make this proposal deal only
with the problems raised in the bug report, and save other cleanups,
including this one, for later.  The ghost label is caused by a failed
attempt to delete the reference to the requirements from the normative
referneces and move it to the non-normative references.  I expect it
will be fixed in the follow-on editorial propsal.

Comment 10 David Ezell 2009-04-17 16:41:11 UTC
6553: XML 1.0 and 1.1 references are inconsistent
http://www.w3.org/XML/Group/2004/06/xmlschema-1/structures.b6553.html
http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b6553.html

To WG 12 April 2009.

6553 XML 1.0 and 1.1 references are inconsistent

Summary: WG decided to: (1) Use undated links, not dated links,
to W3C specs.  (2) Give document status and date ("W3C Working
Draft, 35 March 2013").  (3) Add text to normative references to
W3C specs saying "The edition cited is the one current at the
date of publication of this specification. Implementations may
follow the edition cited and/or any later edition(s); it is
implementation-defined which."

Note: Proposal does NOT make the sentence in (3) a blanket
statement but added to individual references; should be a case by
case decision.  Did NOT add it to non-normative references.  Did
NOT apply decision to submissions (which are not W3C documents.
Did NOT fix other problems in the references.

Some discussion in Bugzilla; amendments needed.

MSM's recommendation:  quick.

   - Mark various undated links as additions, as specified in
     comment 7.

   - Correct the URI for XPointer as specified in comment 8

(see ammendment "and/or"
Comment 11 C. M. Sperberg-McQueen 2009-04-21 18:06:02 UTC
The proposal mentioned in comment 6 was adopted by the XML Schema WG
on its call of 17 April, with amendments; the changes have now been integrated
into the status-quo document.

Henry, as the originator of the issue, please CLOSE or REOPEN the issue to
signal that you are, or are not, content with the WG's resolution of the
issue.  If we don't hear otherwise from you by Friday, we'll assume you are
dancing cartwheels in the atrium of the informatics building to express
your satisfaction with the outcome.