Public Comment

[comment-form@oasis-open.org]

Comment from: dennis.hamilton@acm.org

Name: Dennis E. Hamilton
Title: System Architect
Organization: NuovoDoc
Regarding Specification: OpenDocument 1.0

[I am studying the specification to understand the constraints and integration case for the OpenFormula work that David Wheeler is spear-heading.  The lines of these messages don't wordwrap on display because I am using the comment form provided on the OpenDocument TC page.  I use that form to be clear that I am affirming my agreement to the terms and conditions on that form page.  I haven't tried posting directly to the list by e-mail.]

On p.176, section 8.1, Basic Table Model, the 3rd paragraph says:

"Table rows may be empty, and different rows
might contain a different number of table cells.
This is not an error, but applications might 
resolve this in different ways."

The section goes on to say "Spreadsheet applications *typically* ..." (my emphasis).

I don't find that particularly normative, and I have to assume that, absent MAY, SHOULD, MUST, and other [RFC2119] language, it isn't.  Maybe TYPICALLY should be made a term of art here.  (Maybe the TYPICALLY material should be in an Appendix of non-normative guidance, too.)

I am raising this because it causes difficulties at the edges for integration with other features.   In OpenFormula, there is the quandary about what to say when there is a reference to a cell that, in one sense or another is not there, or may be virtually there. It doesn't seem that there is any way to talk about simply by reference to concepts in the OpenDocument specification.  

I am not sure how it is possible to proceed in the face of this kind of language.  I am confident that OpenFormula will proceed, but it would be great to have guidance with respect to this OpenDocument case, and this "typically" case in particular.

 - Dennis