Re: [office] Formula Sub Committee Draft Charter

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

Michael Brauer - Sun Germany - ham02 - Hamburg wrote:

> David,
> as you probably have seen, the objective of the proposed formula sub 
> committee charter is to develop a specification for spreadsheet 
> formulas that may become part of the OpenDocument specification. This 
> means, a specification that is owned and maintained by OASIS. In 
> addition to this, no decission has been made whether the development 
> of this specification will be based on one or more proposals that TC 
> members may contribute to OASIS, or will be developed from scratch.

Yes, that I understand all that.

Since the formula discussion has been delayed (to deal with IRIs),
now is a good time to discuss the spreadsheet 
formulas issue.  I think we actually all have the 
same basic goals; we just need to work out how to make it happen.  The 
main issue, to me, is to just close a few legal loopholes.
Once that's straightened out, I think that things should go very well.

> From my understanding, your are working on a formula specification 
> (OpenFormula) that you have not contributed to the OASIS OpenDocument TC.

That is correct, but that was also true of OpenDocument.  OpenFormula's 
current state is the same as the original OpenDocument specification; it 
is being developed so it can be submitted as a base document.

> To be honest, I think this may cause a conflict of interests for you 
> if the TC nominates you as SC chair. To help the TC to understand 
> whether such a conflict of interests may exist, I think it would be 
> very useful if you could explain your plans regarding the OpenFormula 
> specification and project to the TC.

Actually, I do not think there are any conflicts of interest.
I think we all have the same objective. I've been working on solving 
the problem of defining formulas in OpenDocument for over a year,
and OASIS is now declaring that it needs to work on a solution.
Sounds like a perfect match!

As far as "plans for the project", the whole point of the project
was to create a base document for a standards body, e.g., OASIS,
that could be submitted to it.
Standards bodies typically start by using a base document; at the
time there were no real base documents. Now, thanks to the OpenFormula
project, there _IS_ a base document available to OASIS.
At the time I was not part of the OpenDocument TC, so I could
not propose or work inside the TC more directly.

I see no conflict in turning to the person who believed there was
a problem, and worked for free to help you solve it even before you asked!
There are no competing base documents available to the TC.
Also, I represent no particular spreadsheet implementation, so
I'm completely free from that particular perceived conflict of interest.

However, it's always fair to ask if there are conflicts of 
interest.  Hopefully, I can ease your mind with my replies. If not, 
let's find out what the concerns are and address them.  I believe that
we all agree on the basic goal, and so it should 
be merely a matter of working out the mechanics.  Basically, I would 
like some legal loopholes eliminated, to make sure that the goal that 
everyone ALREADY shares is committed in writing.

> I think the TC is in particular interested to know whether you will 
> contribute the OpenFormula specification to the OpenDocument TC, and 
> whether you will continue the OpenFormula project when the formula sub 
> committee has been founded. The answer to the later questions may be 
> different in the cases that the TC decides to use the (contributed) 
> OpenFormula as basis for its works, and for the case it does not, so I 
> would like to ask you to provide anserws for both cases.

It has ALWAYS been my intention to contribute the OpenFormula 
specification to a standards body such as OASIS. Absolutely!
After it became clear that the lack of a formula specification
was becoming a very serious concern with OpenDocument, I
went ahead and started work in the hopes that by doing 
so I could help the progress of open standard standards for
document exchange.

After submission I'm open to all sorts of transition approaches. But the
whole point of OpenFormula was to create a base document for a standards
body, not to live forever in its own little world.

I have not heard of ANY serious TECHNICAL objections to OpenFormula.  
There are no competing formula specifications at this time.
And we're way ahead of anyone who would start now; we 
already have a large pool of research, many people who work with many 
different spreadsheet implementations, and a draft with real content.
I know of no one else who's examined multiple implementations and
then developed a specification for recalculated formulas.
Already one OpenDocument implementation (KOffice) is _explicitly_
modifying their implementation to improve their OpenFormula conformance
(we only have a draft, but they are finding it very useful already).
The spec's not done, but that's strong evidence that we have a good start.
Several OASIS TC members have privately commented that OpenFormula would 
be a very good place to start.  I'm unaware of any fundamental technical 
flaws in its basic approach, and since it's a draft, changes can be made
if (when?) problems are found as part of the normal OASIS processes.

But even if OpenFormula were rejected by OASIS for technical reasons (!?), 
and people wanted to start over, I'd still be happy to lead the group.  
I have a lot of experience now in specifying spreadsheet formula languages
that will be difficult to reproduce.  My goal is a specification of formulas
for OpenDocument; OpenFormula is merely a project to "jump start" the
process, to help you get there.

The fundamental problem in my mind is intellectual rights, not 
technology.  I will GLADLY give away OpenFormula to OASIS,
at no cost, and work to help create a good formula specification.
I have made efforts to ensure that I have the right to make this grant.

All I want is that OASIS commit, in writing, to permanently
making the result an open standard including its descendants.
Details are at: http://www.openformula.org/Rules_for_Contributors
But here's the gist: it must be a specification
that is implementable by /ALL/ proprietary and open source software
projects, from Microsoft Office to LGPL and GPL software.

Based on conversations with OASIS leadership, it appears that
the current OASIS rules can't guarantee this.
For example, it looks like the TC could be disbanded and
re-chartered, and suddenly the Royalty-Free clauses would 
not apply to its successor. Or future versions might be encumbered by 
patents. Or the test suite might cost $200K, which only large vendors 
could afford (OSS projects can often use only freely-available
test suites as a practical matter).

I have been talking with the president of OASIS, OASIS general council, 
Eduardo Gutentag and others. As best I understand it, I think
we all agree that these are reasonable goals (though I haven't had much of
a chance to talk with Eduardo yet). We just haven't found a wording that can
guarantee those things. I don't think the legal loopholes were intended.
All I want is to ensure that the legal loopholes are closed.
In the long term I think these loopholes need to be fixed in
OASIS' IPR policies, but in the short term signed letters and
other things should be sufficient to resolve this.  Sun did the same
thing to kick off OpenDocument, a precedence that suggests we can
basically use the same approach again.

--- David A. Wheeler