PLEASE PROOF: The draft minutes from today's WG4 teleconference are at http://www.itscj.ipsj.or.jp/sc34/wg4/def/sc34-wg4-2013-0250.zip

[John Haug]

Re: my action item: John H. to investigate whether an attribute in an ignorable and non-understood namespace should be ignored on an extension element in a namespace that is non-ignorable.  I found that I did indeed investigate that, way back just after the previous call when the question came up.  I didn't see my notes when we got to it during this call.  My apologies!

A reminder of the issue...
The first paragraph of Part 3 includes: "Extension elements suspend Markup Compatibility processing within their content."  This basically sets up a boundary inside of which MCE state (what's ignorable, etc.) is reset, as I understand it.  Is the intent that the boundary starts with the extension element or with its children?
 
e.g., if the ext element has attributes in a non-understood namespace previously declared as ignorable and the ext element is in a different (understood) namespace.  In the following, is a:attr2 ignored if a is a non-understood namespace?
<foo mce:Ignorable="a">
  <extension attr1="..." a:attr2="...">
    ...
  </a>

I checked with one of our developers here and his position is that the package-level processing code in Office would ignore a:attr2.  That level of the implementation - that is, not WML/SML/etc.-specific - doesn't know anything about extensions and would apply the existing MCE state to the extension element.  That extension tag would be handed off to the ML-specific code (e.g., Word, Excel, PPT) to determine whether the attribute is understood.  The scenario above is that it's not understood, so the general package code would be told it's not understood and what happens with the extension is implementation-specific.  In our case, the ML-specific code would have the opportunity to note the raw XML if it wanted to round-trip that blob of XML (save/retain it in the file when it's saved later).

Leaving aside that particular implementation of one codebase that separates the OPC/MCE from the Part 1 and 4 MLs while providing communication between them, in our WG 4 parlance of a unified configuration and codebase, the a: namespace is not understood but we can still say the outer level of an extension declaration is processed using the current MCE state.

John

-----Original Message-----
From: Rex Jaeschke [mailto:rex at RexJaeschke.com] 
Sent: Wednesday, January 9, 2013 3:29 PM
To: 'SC 34 WG4'
Cc: 'TC45'
Subject: PLEASE PROOF: The draft minutes from today's WG4 teleconference are at http://www.itscj.ipsj.or.jp/sc34/wg4/def/sc34-wg4-2013-0250.zip

DR 13-0001, "DML: Theme part root element is incorrect": We agreed to move this to Last Call, as proposed in the latest DR log.

Rex