DR 13:0013: Proposed solution to "the yellow ones"

[Arms, Caroline]

For the "rogue" attribute, is there any acceptable way to recommend in the prose that it not be omitted either for this particular attribute or more generally for cases where an implementation will have to assume something when displaying a document.

  Just a thought,   Caroline

-----Original Message-----
From: Chris Rae [mailto:Chris.Rae at microsoft.com] 
Sent: Tuesday, July 29, 2014 6:00 PM
To: e-SC34-WG4 at ecma-international.org
Subject: DR 13:0013: Proposed solution to "the yellow ones"

I know we've still got some iteration to be done on the "purple ones", but here's the next installment. The subsection of the DR (https://onedrive.live.com/view.aspx/Public%20Documents/2013/DR-13-0013.docx) that this addresses is attributes where defaults were missing in schema and could be just added once we ascertained what those defaults might be.

To determine the defaults I have consulted developers on Microsoft Office and LibreOffice. In all but one case, either the defaults are the same or the feature isn't yet implemented in LibreOffice. In the case where LibreOffice uses a different default, I'm proposing a change to the prose specifying an implementation-defined default. I'd be interested in any other ideas about how to deal with this - I notice in my testing that both LibreOffice and MS Office always seem to write the attribute (which is likely why this has never come up as a practical interoperability issue between suites) but unfortunately I think making the attribute mandatory runs too high a risk of breaking other existing implementations.

Attached are the proposed changes, and also a sample file containing a table with unspecified size on the top border (the rogue attribute in question), in case anyone else wants to use this sample for testing purposes.

Chris