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

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

> I note that in the formula spec you have defined a "PortableIdentifier" 
> that looks like an XML name to me. In other words, it would qualify as 
> an XML ID.
> 
> Is there some reason to not restrict the range of names for expressions?

I presume you mean "NamedExpression", which is the name of syntactic
unit that refers to a named expression.  The reason at the time was
_flexibility_. The current approach defines a way to exchange named
expressions which have _arbitrary_ names, including ones with spaces, etc.
At the time, the thinking was "better to have a way to exchange anything".

I still think that'd be useful to have a way of referring to arbitrary
names, for arbitrary uses of OpenFormula. After all, it
which might not be in a spreadsheet, or even in an OpenDocument file.
But we could certainly entertain the idea of a more restricted set of
names that are permitted in spreadsheets.

"PortableIdentifier" is going after a different issue.  It is merely a
_should_ for portable documents, but it's a useful should.
The problem is that some spreadsheet programs have trouble with
names that look like cell addresses, like "A1".  Traditionally, they've
just assumed that the number of columns supported by spreadsheets
will never increase, but of course, this has been a completely false assumption.
Many users _do_ use names like "QTR1".  So it's suggested that
document creators should try to shy away from them... but we make
sure that we provide a way to handle them anyway.

Now, there's an additional challenge: Scoping.
xml:id values must be unique for an _entire_ document.  However, 
named expressions can be local to a sheet, or global to an entire table.
So there can be _different_ expressions with the _same_ name
inside a document (e.g., one with multiple tables or multiple sheets).
Also, it's perfectly fine to have a name of a named expression that is
also the name of a style or other things you might also want to identify with
xml:id.  So you'd have to prefix the xml:id in some way to handle all that.

So to change, we would need to:
1. Determine that limiting named expression names to xml id's is
okay.  This would mean that we couldn't have spaces, for example.
2. Determine how to prefix/affix names to make them truly unique
in a document.

--- David A. Wheeler