Building collection loop efficiency

[snow-raven]

General looping efficiency question for building a collection of controls;

Do you think I'd be more efficient evaluating all of the property tags that I want to add to a collection before moving to the next collection, or would it be more efficient evaluating each tag & adding it to all of the appropriate collections before moving to the next tag?

Example:

For Each ctl In Me.Controls
    If ctl.Tag = "Project" Or ctl.Tag = "Intervals" Or ctl.Tag = "Lab" Or ctl.Tag = "TestSetUp" (etc...) Then
        colctlsAll.Add ctl, ctl.Name
    End If
    If ctl.Tag = "Collar" Or ctl.Tag = "Intervals" Or ctl.Tag = "Lab" Or ctl.Tag = "TestSetUp" (etc...) Then
        colctlsNewPrjt.Add ctl, ctl.Name
    End If
    If ctl.Tag = "Intervals" Or ctl.Tag = "Lab" Or ctl.Tag = "TestSetUp" (etc...) Then
        colctlsNoCllr.Add ctl, ctl.Name
    End If
Next ctl
Set ctl = Nothing

Or:

For Each ctl In Me.Controls
    If ctl.Tag = "Project" Then
        colctlsAll.Add ctl, ctl.Name
    End If
    If ctl.Tag = "Collar" Then
        colctlsAll.Add ctl, ctl.Name
        colctlsNewPrjt.Add ctl, ctl.Name
    End If
    If ctl.Tag = "Intervals" Then
        colctlsAll.Add ctl, ctl.Name
        colctlsNewPrjt.Add ctl, ctl.Name
        colctlsNoCllr.Add ctl, ctl.Name
    End If
Next ctl
Set ctl = Nothing

 

[June7]

Time the procedures and find out which is faster.

dteStart = Now()
'run procecure
Debug.Print DateDiff("s", dteStart, Now())

 

[theDBguy]

Also, wouldn’t it be simpler to use a Select Case statement?

 

[The_Doc_Man]

I'm of the opinion that you should make the changes to one collection at a time, though it MIGHT be a marginal speed difference.

I say that because of the "collection probes" required. In essence, each collection can be thought of as an array or list. A collection that contains collections is a list that contains lists. It is a nested, double-barrelled search.

If you do the individual INNER additions one at a time, you are repeatedly searching the OUTER collection for a match, then searching the INNER collection. If you instead do the OUTER collection search and, when you find a match, do all INNER items at once (while keeping track of the current collection object, i.e. NOT searching for it - 'cause you found it), I think you do less searching, all other steps being equal.

One man's opinion.

 

[MajP]

No difference.

 

[The_Doc_Man]

MajP says NO DIFFERENCE, I say MARGINAL DIFFERENCE AT BEST. So in the end analysis, you make your choice and take your chances.

 

[snow-raven]

Thanks, all. I don't know why I didn't think of applying a timer. I'll try that when I get a second. (hyuk, hyuk)

Doc_Man, I did nearly implement a select case, but I couldn't think of a way to apply it that didn't include it in each If statement, so it didn't seem that it would add much to efficiency. Admittedly, I have even less experience using select case than with loops in general.

Thanks for the feedback, everyone!

 

[MajP]

IMO waste of time. So lets say technique A is 75 milliseconds faster 65% of the time? So what. I severely doubt you can measure this with a timer, you will need something far more advanced and any measure will be noise.

 

[NauticalGent]

Sounds like a job for Col (Isladogs)...

 

[The_Doc_Man]

Like I said, MajP, minimal difference. What was it that P.T. Barnum's barkers used to say in the side-show carnival? "You pays your money and you takes your chances."

 

[isladogs]

Sounds like a job for Col (Isladogs)...

Sorry but I'm going to opt out of this one
I also see little difference being likely.

Whilst you can easily measure to centisecond (0.01s) accuracy, doing anything more precise than that needs more complex code.