New version, PLEASE LOAD IT BEFORE CONTINUING

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

You'll notice that I've just posted an updated version of the document.
PLEASE LOAD IT BEFORE EDITING ANYTHING, because I've modified
OTHER'S SECTIONS and I really don't want my edits lost :-(. It's at:
 http://www.oasis-open.org/apps/org/workgroup/office-formula/download.php/19898/openformula-20060823.odt

Indeed, for anyone editing the file, please reload the Wednesday evening
version before continuing.  This merge/resend gives me a short time
to make structural changes or fix stuff that's group-wide.  For example,
I've moved the rounding functions (ugh!) to their own section; they are
so bizarre, and make sense to consider together, that at least for now
I think this will help us.
Also, I added a number of more rigorous mathematical definitions
in various sections, thanks to Zwillinger's "CRC Standard Mathematical Tables
and Formulae", which is also added to the references.  Also added various
test cases, information, etc.

Let's welcome aboard Frank Stecher, who works at Sun on spreadsheet
functional QA.  Frank is going through the paperwork process, but I expect
he will be formally listed soon.  In the meantime, we'll put him to work :-).

Here are the current "locked" sections.  Please send in your update
Aug 30, Wednesday 7PM EST (whatever you've completed so far).
You'll notice that the financial and statistical sections are subdivided -
we have a number of undefined functions in those sections, so having
multiple people work them is a good idea (methinks). Assignments are:
    * Andreas J. Guelzow - statistical (S-Z)
    * Tomas Mecir - statistical (A-J)
    * Frank Stecher - statistical (K-R)
    * Robert Weir - financial (A-E)
    * (Helen) Zhi Yu Yue - financial (N-Z)
    * Bryce L Nordgren - bit operations
If we get someone else new, or someone completes early,
the next available section to assign is Financial (F-M). 

As always, these assignments and instructions on what to do are at:
http://wiki.oasis-open.org/office/Formula_Work

And ANYONE on this list is a reviewer, so please post issues and/or
send email to the person with the lock, so that they can fix it.

I've tweaked the instructions on what to do so that they talk
more about how to create good test cases.  We want at least 9 digits
of precision where that's reasonable to provide, though if you're using
epsilon only 6 digits matter for now.  But who knows, we might want to
make epsilon a little smaller, and having extra digits makes that easy.
But only include them if they're right :-).  And not every function needs to be that
precise; we just need SOME range so that we eliminate false failures.

Besides filling in the undefined functions, handling the remaining TODOs
(inc. handling the comments on syntax), and reviewing it all so that
we have a high-quality result, I think there's
one other thing to do: review the literature for known bugs in spreadsheet applications,
and add test cases for them. I've begun collecting pointers to such data here:
 http://wiki.oasis-open.org/office/Spreadsheet_Bugs
If you know of any spreadsheet bugs, please add pointers to it on that page.
Even if it's only historical, if it happened before it could happen again.
The whole goal is to make sure that the same formula produces the
"same" answer everywhere, and obviously that can't happen if somebody
has a bug.  We cannot test for all possible bugs, but I think we need to
consider this data source as an important source.  I'd love to assign someone
the job of "search for bugs, and create test cases with the CORRECT result";
volunteers?  Perhaps that's something a new person could do for us.

We seem to get a lot of FUD.  If you hear FUD, please point people to:
 http://wiki.oasis-open.org/office/About_OpenFormula
If there's a new FUD issue, please add the counter to that section.

--- David A. Wheeler