Re: References to XML in SML specs

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

John Arwe writes:

> (a) It was noted that the language proposed to the AC rep could be read to 
> mean that ANY version of XML is allowed; thus impl1 using 4th edition only 
> and impl2 using 5th edition only would be within the bounds of the 
> proposed language.

Correct.  10 years from now requiring 4e support will look silly.

> (b) The working group has had a fairly long-standing goal that the base 
> level of the specs (MUSTs, floors) together serve as an entry point: a 
> minimum level such that conformant documents and implementations would 
> enjoy "guaranteed" interoperability ("wide" perhaps being more accurate 
> for the language lawyers).  The interpretation in (a) conflicts with this 
> goal.

5e is backwards-compatible.  Any 4e-well-formed document is
5e-well-formed.  So for documents there is no problem, and indeed no
point in requiring 4e support, since a 5e-only processor will always
work with a 4e-compliant document.

And (still digging :-) any interop problem which might arise under the
proposed wording will also arise under a (MUST support 4e, MAY support
e>4).  The problem scenario, which can arise in _either_ case, is that
someone using a 5e-supporting implementation writes an SML document
which uses 5e-only names, because it governs documents with elements
which have such names.  Such an SML document will not be accepted by a
4e-only processor.

The bottom line is that people in Japan, India and the Middle East
will be writing XML documents with 5e-only names.  Either SML
implementors will allow such documents to be processed, or they
won't.  My proposed wording _allows_ conformant implementations to
process such documents.  It also _allows_ conformant implementations
to reject such documents as ill-formed.

> (c) Language was proposed to clarify our goal (require 4th edition, allow 
> all others), and is in the minutes.  The wg believes this to be consistent 
> with the (the wg's intended, at least) spirit of the earlier proposals.

I would prefer the wording I proposed, for the reasons given above.
In practice I don't believe the new wording is actually coherent.
What does it mean for a processor to 'support' both 4e and 5e?  That
it both rejects and does not reject a document with 5e-only names?

> (g) It was noted that a number of recent Rec-track documents, including 
> recent Rec's, do not appear to have their references "in good order", i.e. 
> if SML simply followed their lead it appears likely that the respondent's 
> (SML) comment would not be addressed.  That results/ed in confusion, and 
> to some degree a sense that the "best practice" in this area is 
> ill-defined.  I took an action item to open a bug against Schema 1.1 on 
> this issue, since that was one of the examples we consulted for best 
> practice (sorry, I know you're an editor there :-) .

Please see the agreed resolution of an existing bug "XML 1.0 and 1.1
references are inconsisten[t]" [3].  That's where the wording I
proposed came from.  Indeed there are other specs which are broken in
this regard, we're doing our best to get them fixed as well. . .

> (h) There was/is a certain fear over the ramifications of fixing more than 
> just the XML issue, wrt whether or not doing so would be considered 
> substantive [2] and the implications if it were considered
> substantive.

I can reassure you that as long as 4e-only is allowed, there will not
be any complaint of substantive change from the Director.

ht

> [1] http://lists.w3.org/Archives/Public/public-sml/2009Apr/att-0011/20090406-sml-minutes.html#item05
> [2] http://www.w3.org/2005/10/Process-20051014/tr.html#transition-reqs
[3] http://www.w3.org/Bugs/Public/show_bug.cgi?id=6553
- -- 
       Henry S. Thompson, School of Informatics, University of Edinburgh
                         Half-time member of W3C Team
      10 Crichton Street, Edinburgh EH8 9AB, SCOTLAND -- (44) 131 650-4440
                Fax: (44) 131 651-1426, e-mail: ht@inf.ed.ac.uk
                       URL: http://www.ltg.ed.ac.uk/~ht/
[mail really from me _always_ has this .sig -- mail without it is forged spam]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFJ3JJCkjnJixAXWBoRAlEkAJsF4Y0K3vLoLJsjwBLc+vM//ll/kgCfbNHO
Elrm+rbuxDVj6ou/CYZGd1U=
=AEiA
-----END PGP SIGNATURE-----

Received on Wednesday, 8 April 2009 12:02:55 UTC