RE: [office] Errata: Substantive Schema Change in 15.27.22?

["Dennis E. Hamilton" <dennis.hamilton@acm.org>]

While we are thinking this over prior to the next call, I have some further
observations:

1. The only schema file for ODF 1.0 on the OASIS site has the -treshold
spelling (although the 1.1 schema has -threshold).  This is not normative,
but it is something to keep in mind.

2. We do not know if translations of the specification carry the -treshold
and -threshold spellings in literal attribute names and translate otherwise
when threshold is used in the title and in the prose.  So a developer
concluding there is a misspelling in the schema may be a little less
obvious.

3. If no ODF 1.0 implementation has ever supported style:wrap="dynamic" we
would be off the hook.  The one problem is with ODF 1.0 implementations
evidently still being provided and used in order to be IS 26300 compliant. 

 - Dennis

PS: I agree that the easiest way out of this situation is if we could simply
make the correction in the ODF 1.0 specifications.  Other resolutions (since
it is changed in 1.1 and we expect that change to continue into 1.2) are far
messier (unless the feature is still not implemented).

-----Original Message-----
From: Michael.Brauer@Sun.COM [mailto:Michael.Brauer@Sun.COM] 
http://lists.oasis-open.org/archives/office/200809/msg00083.html
Sent: Wednesday, September 24, 2008 04:49
To: Doug Mahugh
Cc: Andreas J. Guelzow; office@lists.oasis-open.org
Subject: Re: [office] Errata: Substantive Schema Change in 15.27.22?

[ ... ]

The section in question reads as follows:

*****
Dynamic Wrap Threshold

The style:wrap-dynamic-threshold attribute is evaluated only if the
style:wrap attribute has a value of dynamic. It specifies the minimum
distance between the page or column border and the object for which
wrapping will be enabled.
<define name="style-graphic-properties-attlist" combine="interleave">
	<optional>
		<attribute name="style:wrap-dynamic-treshold">
			<ref name="nonNegativeLength"/>
		</attribute>
	</optional>
</define>
******

The error report in question is:

******
The first para in 15.27.22 says style:wrap-dynamic-threshold, while
the schema fragment in this subsection says style:wrap-dynamic-treshold.
("h" between "t" and "r" is missing.)
******

We have spelled the attribute name correctly one time (in the
description), and one time incorrectly (in the schema). The heading also
says "Threshold". This means that the spelling in fact is inconsistent,
and the question actually is what implementors reading the specification
would assume what the name of the attribute is. My personal opinion is
that implementors notice the inconsistency, assume that it was not our
intention to misspell the attribute, and implement it with the correct
spelling. I further would assume that they do not implement a misspelled
attribute name (if the spelling is correct in the description) without
notifying us about the issue or asking us what our intention was. I
therefore think that the correction of the misspelled word is not
substantive, even if it occurs in the schema. Please note that my
conclusion would be different if either "treshold" would be a correct
spelled word, or if we would have misspelled it in the description and
the heading, too.

Anyway that is my personal opinion, and I do understand that others come
to a different conclusion.

My suggestion for resolving this issue is that we continue to discuss
this on the mailing list until the end of this week. We may then have a
small ballot in the TC on Monday whether or not to include this
resolution in the errata before we conduct the other three votes
required to start a public review of the draft.

Best regards

Michael

[ ... ]