Re: [office-comment] Conceptual Problem: Ambiguity for Names usedin Formulas

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

Leonard Mada:
> a recent bug-report [1] on the OOo-site points to a more fundamental problem
> of ODF, especially of column naming vs named expressions/ranges.
> [1] http://www.openoffice.org/issues/show_bug.cgi?id=90230
> 
> The problem is not bound to the OpenOffice implementation,
> though I will describe it based on OOo:
> 
> OOo 2.4 currently supports 65535 columns, therefore:
> 'ST1' is a valid name (expression, range, ...).
> 
> However, with OOo 3.0, this limit will be moved upward,
> so that 'ST1' will be now a valid column [cell].
> 
> Ad extremum, IF we consider a very distant future (why not in 10,000 years ;-) ),
> then any string used currently as name will probably point to a valid column.
> [IF only string, then the whole column, IF string_number_ then a cell]
> 
> It is unsafe to assume that the current column limit will be held
> [or any other column limit], and changing this limit in the future
> will jeopardise existing names.
> 
> I propose to enhance the ODF as follows:
>  - names should be marked explicitly as such within formulas
>  - this will allow to unambiguosly identify a name as such in the future
>  - this will be especially improtant, IF the spreadsheet evolves
>    to something basically limtless, when every name will point to
>    a cell/column

Hi - we knew about this potential problem at the beginning of the
definition of OpenFormula, and have already addressed it.

In the exchange format, _ALL_ cell references are surrounded by [...],
while named expressions (variables) are not.  In addition,
to enable handling really odd expression names, you can prefix a named
expression with $$,  Thus, ST1 is a variable, as is $$ST1, while [.ST1]
is a cell reference.  See section 5.8 for cell references, and 5.11
for named references.

OpenFormula doesn't mandate any particular user interface, but
the likely implementation would be that "text that looks like a cell
reference" would be turned into a cell reference, and some indicator
(like $$) would disable that.  The key is that the _exchange_ format
is unambiguous, and works no matter how many rows/columns are allowed.

--- David A. Wheeler