Re: [office-formula] STDEV

Andreas J. Guelzow:
> I would like to propose that we add another test case to STDEV to ensure
> that implementations use a reasonably stable way of calculating this
> expression. (This will make it likely that they use similar code for
> this frequently used expression.) This testcase would need an extension
> to the test data array and be essentially
> 
> STDEV({10000000001,10000000002,10000000003,10000000004,...,10000000010})
> 
> the result should be 3.027650+/-epsilon

Agree.  That's a great idea.

To speed creation, let's just add this with inline arrays.  We can pull out the inline arrays all at once then... it's a pain to do individually.

Who wants to write that up?

--- David A. Wheeler