a compatibility sheet and some results

[Nicholas Lamb <njl ecs soton ac uk>]

http://www.ecs.soton.ac.uk/~njl98r/excel/funa.xls

I've uploaded a spreadsheet which runs some rather more comprehensive tests
on functions starting with 'A'. I selected functions which were documented
in my outdated Excel reference books, thus they have been in use for more
than a decade, and the XLS file was created in the current version.

15 functions, 10 have identical results in Gnumeric the remainder are

ACCRINT fails with some missing parameters, which is arguably fine, except
        for =accrint(33297,33481,33359,0.1,,2) which Excel defines as
        equivalent to =accrint(33297,33481,33359,0.1,1000,2)
        We'd need a securities expert (or interested amateur) to tell us
        if Excel's assumption here is useful, perhaps even expected.

ADDRESS failed because Gnumeric always quotes sheet names

AND fails because Gnumeric considers strings to be TRUE

AREAS fails because Gnumeric returns #VALUE! for an invalid Name, while
      Excel returns #NAME?

ATAN2 fails because Excel defines ATAN2(0,0) as #DIV/0! while Gnumeric
      returns 0.

Jody, if you'd like I can file bugs for some, or all of these depending
on your judgement of whether they're good candidates for improved
compatibility.

I also loaded this sheet up in OO Calc, which fails every function, often
on inconsistent error handling or odd behaviour differences.

This file might move, and will hopefully be joined by more later.

Nick.