Would YOU use it ... or not?

Depends on what I was using it for. There's really no inconsistency in
DATEDIF. It works absolutely consistently, and if you know how it works,
you can absolutely predict the results. The problem is that "month" is
used inconsistently in problem statements.

To get

X years, Y months, Z days

where Z is calculated as =DATEDIF(date1,date2,"md"), definitely not. To
use the "y", "m", or "yd" switch, perhaps.

OTOH, there's absolutely no function that can provide a general solution
to the problem of the above format for, say,

date1 = 31 January 2005
date2 = 1 March 2005

Since "months" is an inherently fuzzy concept, the number of months and
days is also fuzzy, and has more than one valid solution. If one chooses
a particular definition of month, then a consistent algorithm can be
developed.

If this is to be used for decision making (e.g., promotions, layoffs,
etc.), it's absolutely critical that you use an algorithm that matches
the HR policy at the plants. How do *they* define months and days of
service? Make your algorithm (whether using formulas or a UDF) conform
exactly to their specifications.

In article ,
"RagDyeR" wrote:

About to start out on a seniority project for the plants.

If this was 3 months ago, I wouldn't have given a second thought about using
Datedif.

However, in the past couple of weeks, I've read, and been part of threads
where the inconsistencies of the function have been brought to light.

Of course, seniority will range over the full gamut of time, from days to
years.

Would you please just share with me your opinion on the function?

Would YOU use it ... or not?