Re: Please participate in OpenFormula! First issue: Where should the mailing list be?

On Wed, Oct 19, 2005 at 10:42:32AM +0200, Niklas Nebel wrote:
The order of a function's parameters would need to be 
specified, but not the details about its results. Just like in a word 
processing document, where the file format says nothing about where on a 
page a single word will appear. It only stores the logical structure. 
For interoperability, similar-looking layout results are important, but 
that's a separate issue, not part of the file format.

This seems to be the fundamental point of contention.  I don't see
the results of the calculations as being something that can diverge
without degrading interoperability.  We've already split this hair
by having OpenDoc vs OpenFormula.  To split it further seems counter
productive to me.

If there is a requirement of "the same results", that doesn't leave much 
room for calling things like non-existing dates, rounding in value 
comparisons, or conversion from text or boolean to numbers 
"implementation specific".

The requirement is being driven by the uncountable number of
existing spreadsheets that we need to support. 
In fact, a specification that says "the same results as Excel" doesn't 
add much value at all. Today, Excel interoperability is already a goal, 
but complete compatibility isn't reached by any application. With such a 
specification, OpenFormula compliance will be a goal that nobody 
completely achieves. So what is gained?

Beyond the obvious benefits of providing a shared document for
spreadsheet implementors to explain the required behaviours, this
spec would give us the focus to fix our applications to conform
rather than trying to do one of patches at the import/export layer.

If we include a validation suite with the specification we end up
with several powerful tools.
    - A metric to measure compliance.
    - A shared set of regression tests
    - A way to begin taking control over the evaluation semantics
      the same way Daniel Rentz's xls file format docs give us some
      control over the xls format.

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]