Comments on: Explicit Conversion Functions

By: emma

emma — Tue, 11 Jan 2005 12:31:00 +0000

Just to add to the 1st 2 comments up here – the rounding “convention” of rounding up at 0.5 is used by the majority of people but when you work in certain industries this is not accurate enough and therefore rounding to the nearest even number is much better – this includes banking (I think this method of rounding has a name something to do with accounting) and forecasting in virtually any industry. Rounding to the nearest even number is actually a “convention” it’s just that it’s only one that’s used where huge amounts of data are found.
The reason that some of the conversions use one convention and some another is so that you can use excel to do any of them.

By: Dick

Dick — Fri, 04 Jun 2004 19:10:00 +0000

jwolf: I’ve read your comment three times and it’s still over my head. Can you give me a short example of where this fails? I’d like to learn what you’re talking about.

By: Dick

Dick — Fri, 04 Jun 2004 19:08:00 +0000

Good point, Juan. Harald convinced me that I should learn .net, now I just have to do it.

By: jwolf

jwolf — Fri, 04 Jun 2004 18:07:00 +0000

How about CDec when filling a variant array with mixed data vectors? Keeps XL VBA from auto-assigning as literal datetime when reading first cell(datetime) into a vector and preserves for calculations in VBA.

By: Juan Pablo

Juan Pablo — Fri, 04 Jun 2004 18:00:00 +0000

One good thing about this explicit conversions (altough I never use them either ), is that it makes the transition to .Net easier…

For example, this

Dim i As Integer

i = 1

If i Then
‘Something
End If

would work in VBA, but would cause a compile error in .Net.

By: Dick

Dick — Fri, 04 Jun 2004 17:42:00 +0000

vkd: That makes sense. My problem is that we have a convention and this doesn’t follow it. Even if it’s arbitrary, it’s what I would expect (which I guess is why it’s a convention – to meet expectations.) If CLng and CInt break convention, then why doesn’t Round?

By: vkd

vkd — Fri, 04 Jun 2004 16:10:00 +0000

You Write:
" According to help, CLng rounds decimals that are exactly .5 to the nearest even number. There's probably some lucid explanation for that, but it just seems crazy to me."

Its not crazy. Consider that values ending in .5 are exactly between the nearest integers. The convention of always rounding them up is arbitrary, and can infact introduce a positive skew into a very (very) large data set. In a really big data set, rounding x.5 to the nearest even will round down as often as it rounds up, and thus eliminate the potential skew.