robert_weir@us.ibm.com wrote:
> > I received a request from a colleague on the Accessibility
> > Subcommittee related to nested tables. The request is for us to
> > deprecate the use of the is-substable attribute in section 8.2.6.
> > This attribute does not work well with screen readers and the
> > enhanced support we have in ODF 1.1 and 1.2 is preferred for nested
> > tables.
> > A more detailed analysis of the issue is here:
> > http://lists.oasis-open.org/archives/office-accessibility/200608/msg00048.html
I do NOT agree that "is-subtable" should be deprecated. I think that subtables are an important GOOD feature of ODF. Instead, the problem appears to be that some ODF implementations do not provide sufficient control and discoverability of subtables when contrasted with cell spanning. In other words, this is an implementation user interface issue, not a problem with the exchange format itself. I suggest that the accessibility group work with implementors to find a better way to control and reveal subtables vs. spanning, and then document best practices for implementation user interfaces.
Why should subtables continue to exist? Here are a few reasons:
* Subtables are necessary for capturing the correct semantics of many existing documents, including those written in HTML and Docbook. HTML supports BOTH spanning AND subtables, and yet in general is a weak format at capturing semantics. A document format that can't even capture HTML correctly is REALLY weak. Docbook also supports both spanning AND tables in tables (see EntryTbl). OpenDocument should tend to support a _SUPERSET_ of the capabilities of legacy document formats, not a subset.
* MANY existing OpenDocument tables use subtables. They have been there since the beginning. Losing this capability will mean the loss of ability to read or edit many existing files.
* As practical matter, subtables are really useful for expressing formats and their relationships and are NOT simply "another way to do spans". If you never edit a table, you can sortof work around the lack of subtables, but as soon as you start editing a table, the differences of spanning vs. subtables immediately come to light. Having subtables AND spanning is very useful for expressing a table's semantic structure, which then translates cleanly into its presentation. Which is why it's in HTML, Docbook, and other formats in the first place.
* The lack of subtable support is a known serious weakness in Microsoft XML, and is one of the main things that various parties are appealing to Microsoft to fix. It'd be horrific if OpenDocument _dropped_ one of its advantages, while a proprietary format _added_ that capability.
Andreas said:
> I think the real problem is that some word processor developers are
> using these sub tables for things that semantically are not subtables.
> This is little reason to weaken the odf format itself.
Agreed. If a user commands a program to "span" a cell over multiple cells, it should be saving that as a span, while if a user commands a program to "insert a table here" inside a cell, it should insert a new table. In addition, there should be some sort of status display that shows the user where they are, one that is fully accessible (e.g., so the blind can know their current "location" in a table or subtable as they move into it). I can see the accessibility document giving recommendations to implementors along these lines, recommending how to control and expose status so that blind users (in particular) can deal with them.
Removing a powerful and useful utility from both the seeing and the blind hurts both. Instead, let's make sure that both the seeing and the blind have full access (!) to powerful capabilities, ones that ALREADY exist in OpenDocument.
--- David A. Wheeler