Re: Gnumeric 1.0.3

[Nick Lamb <njl98r ecs soton ac uk>]

On Tue, Jan 22, 2002 at 02:27:52AM -0500, Jody Goldberg wrote:
> Gnumeric 1.0.3 is now available.

This meant (because I don't run Gnumeric from CVS) that I was able to take
5 minutes out this morning to check the COUP*() functions and see how they
compare to those 97coup*.xls and xpcoup*.xls files from Excel

I think (still verifying) that Excel 2000 + Y2K fixes is the same as Excel XP
as far as the COUP*() functions are concerned. Gnumeric seems to agree with
Excel 2K / XP about coupncd() coupnum() and couppcd() completely too.

However there are still some differences between Excel 10 (XP) and Gnumeric
1.0.3 for COUPDAY*() functions, only they are now much rarer. Are these
remaining differences errors in Excel XP? Hopefully our resident financial
analysis experts can tell me. Here are some examples...

coupdays(1996-10-22, 2002-02-08, 4, 1) = 91 in gnumeric, 92 in Excel XP
coupdays(2015-05-29, 2016-10-31, 2, 1) = 184 vs 183

coupdaysnc(1996-10-22, 2002-02-08, 1, 0) = 126 in gnumeric, 128 in Excel XP
coupdaysnc(2006-12-05, 2013-11-28, 4, 0) = 83 vs 85

coupdaysnc(1996-10-22, 2002-02-08, 1, 0) = 233 in gnumeric, 232 in Excel XP
coupdaysnc(1992-08-03, 1996-08-29, 2, 0) = 152 vs 153

[Amusing side note, Microsoft KB article Q282847 seems to imply that all
 the "Analysis ToolPak" features like COUP*() are implemented in VBA.
 No wonder the quality is so poor, performance and correctness are hardly
 the hallmarks of VB programming in general. ]

Nick.