Re: [office] Formula subcommittee status

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

I said:
> > table cell entries end up looking like this:
> >   <table:table-cell table:formula="of:=5+2*3"></table:table-cell>

Bruce D'Arcus said:
> Hmm ... hold on a second:
...
> QNames in attributes are sort of a no-no, aren't they?

No, they're widely used (XSLT 1.0, XML Schemas, Canonical XML, etc.).
It's true that some complain about them.

> <http://www.xml.com/pub/a/2002/02/06/deviant.html>
> This has practical implications beyond formulas, so it might be worth 
> discussing now?

Discuss, sure!  And better now than later, I guess.

But:
* This is already recommended in the official OpenDocument standard.
   Section 6.7.6: "formula should start with a namespace prefix [that]
   indicates the syntax and semantic used within the formula."
   Section 8.1.3: "Every formula should begin with a namespace prefix
   specifying the syntax and semantics used within the formula."
* This is already standard practice.  OpenOffice.org, for example,
   writes all formulas with the "oooc:=..." prefix as the initial text
   of its attribute, with the oooc: namespace defined in the usual way.
* The URL you cite, <http://www.xml.com/pub/a/2002/02/06/deviant.html>,
   is just as much a proof that it's standard practice:
   - Evan Lenz doesn't like them, yet admits that they
   are widespread, noting uses of it include "XSLT 1.0, XML Schemas,
   Canonical XML, and other specs".  He goes on to say
   "I could easily be persuaded... [that] we have to stick with the status quo,
    that it will cause more damage to remove the knife than to leave it in."
   - Jonathan Borden: "...usage of QNames in attribute values
     is a fact of life and has been since XSLT 1.0... Oh well."
   - Ronald Bourret suggested that it's better than the obvious
     alternative, using a URI and a local name. In XML-DBMS, "...when we refer to
     an element type name, we use a QName. We could have used two separate
     values (URI and local name), but if you've ever typed this in more than once,
     you'll quickly realize that QNames are much friendlier..."
* We NEED to be able to distinguish between different formula languages.
   This supports painless upgrades, for example, and that's critically important.
* Frankly, the alternatives I know of look worse.  There is a _reason_ people
   put QNames in attributes: they solve a problem.  Having a separate
   "namespace-naming" attribute, or having a full URL embedded as a prefix in
   each attribute, is far worse.

If there's a better solution, now would be the time to discuss it.
But since this is already standard practice, and recommended by the
official standard, I think we should proceed as-is.  It works.

--- David A. Wheeler