Re: [office-formula] A-functions

[Eike Rathke <erack@sun.com>]

Hi David,

On Thu, Sep 07, 2006 at 10:56:01 -0400, David A. Wheeler wrote:

> > > - OpenOffice2: non-A versions
> > Not exactly. The non-A-versions taking NumberSequences always ignore
> > non-inline text, it is not converted to 0 as may be seen with the
> > AVERAGE function, while inline-text produces an error.
> 
> I don't have Excel handy this instant, but I'm pretty sure taht for it, inline text
> as a parameter is converted to a number (where possible; I suspect it's
> an error if it can't, not sure).  Walkenbach goes into this in great detail for SUM.

Well, yes, for Excel, we were talking about OOo in that case ;-)


> The "A" is really Excel's naming convetion for "these are Lotus 1-2-3's
> semantics", so the "A" doesn't imply equal semantics.  Unfortunately.

Hmm.. I always regarded the A as some 'A'lpha or such, because those *A
functions take strings into account. This may be congruent with Lotus
behavior though.


> > > Also, the functions should accept arbitrary number of parameters, and
> > > automatically expand ranges (so you wan do things like,
> > > AVERAGEA(5.4;788;A1:G8).
> > 
> > This is standard behavior for functions taking NumberSequences.
> 
> AVERAGEA doesn't follow the same rules, and it looks like its funny rules
> are shared by others.  We may need another type (other than NumberSequence)
> to capture its oddities clearly.

True. NonEmptySequence comes to my mind, as empty cells are ignored and
numbers, strings and logicals follow some rules.

  Eike

-- 
Automatic string conversions considered dangerous. They are the GOTO statements
of spreadsheets.  --Robert Weir on the OpenDocument formula subcommittee's list.