>>If I recall correctly, PEMSTATUS works quicker than TYPE. So, if you need it in a loop you may use PEMSTATUS over TYPE (we do need some speed measurements here, I may be wrong).
>>
>But if the n is big, I wonder several calls to PEMSTATUS can still have perfromance advantage over a single call to TYPE.
>

I see now better the point of the original question. I guess it's more theoretical than practical, though, or we're talking about some Treeview structure?

Anyway, we need to test it, I guess for the case like this TYPE may be better (one function call versus N function calls).

>>Also I use the other way around TYPE <> "U". Not sure if there is any difference at all, but in my opinion it's clearer this way.
>>
>em.. I got use to placing constant or manifest constant on the left side in a comparison expression... can't recall since when, something like I read an article which convince me it is better that way.
>Btw, when I search my drive with code reference for codes which I type() test object item, I find myself lucky because I have the above habit.
>I search for "[O] = TYPE(" and I got quite an exact list of what I am interested in.
>
>If I were placing the "U" like you do on the right side of the operator ... then the best way I can think of is to search for "TYPE(" (I wonder code reference supports regular expression...), but then I will have a long list of type() testing for all TYPEs. <g>

Good point! Ok, if that's the case you're certainly correct.