Summary from today's teleconference ("OpenFormula remaining workdiscussion")

["David A. Wheeler" <dwheeler@dwheeler.com>]

We had a fruitful discussion on the informal teleconference today.  Here 
are some of the key take-aways that I got; please let me know what I 
missed or got wrong:

* There was a long discussion about goal and scope of the specification, 
and how to best express requirements.  I'm not sure I can even clearly 
summarize that wide-ranging discussion, but the conclusion is simple 
enough:  I will work with Patrick (and probably Rob Weir) to create some 
draft text to explain more clearly the goal/scope of the specification. 
   We'll then post a rough draft to the mailing list, for comment and 
improvement.  I expect we'll go through several versions of that text, 
and that's fine.  Eventually this will be put in the front of the document.

* There was some discussion about the date system issues:
  - Rob Weir will try to contact a financial expert to help us address 
the date system issues.
  - The bottom line is we need to nail down the various date algorithms 
that people use, and define some reasonable mapping/ID system.  As I 
showed earlier, OOXML and Excel differ (I believe OOXML is in error).
  - I will talk to the Doug (Microsoft TC rep) to see if they'll send us 
an update on the date system material.  Microsoft representatives: If 
you're reading this, consider yourself requested :-).  A while back 
Microsoft helpfully provided two versions of code that were supposed to 
represent "what Excel really did", but they were inconsistent :-(. 
Which obviously made that work hard to build on. Can we get a consistent 
version?  Also, does Microsoft plan to change Excel to match the OOXML 
spec, to submit a change to OOXML to make it match Excel, or neither?
  - I will work with Patrick to start developing a position on what 
should be done with date systems.

* I restated something that more recent attendees may have missed: The 
test cases will NOT be part of the final spec, per a TC ruling.  All 
test cases are marked as "hidden" text; they can be 
displayed/undisplayed at will, and the final version will have the test 
cases stripped away (using an XSLT script Michael B. created).  Having 
the test cases in the draft is useful for our discussions, and it'll be 
very useful if an "annotated" spec is released, but it is NOT going to 
be in the final official version.