Switch or If

[Randall Reetz]

I agree.  Inside a real-time interaction repsonce loop (redrawing a graphic according to mouse loc) is where nested ifs alow minimal interpretor expence.  Also, nested ifs demand a great understanding of the problem topology which may or may not be worth the effort.  Switch statements dont allow easy stacking of decision criteria.  A nested if structure allows hierarchical triage of conditional logic.  A better word for "switch" could have been "first match" as it presents an ordered list of conditional tests at the same level, once a match is found the interpretor exits the switch list.  For performance reasons, the ordering OS switch conditionals should correlate to the natural frequency distribution of your problem domain (most comon conditions at the top).

-----Original Message-----
From: "Phil Davis" <revdev at pdslabs.net>
To: "How to use Revolution" <use-revolution at lists.runrev.com>
Sent: 4/23/2009 8:29 AM
Subject: Re: Switch or If

Hi Camm,

Unless you're building something where the small difference in speed is 
absolutely critical, I would seek to answer a different question:  Which 
structure will be easier to understand and maintain in the future when I 
(or others) come back to it?

My 2 cents' worth -
Phil Davis


Mark Smith wrote:
>
> On 23 Apr 2009, at 11:02, <camm29 at tesco.net> <camm29 at tesco.net> wrote:
>
>> All ,
>>
>> Which is faster in a loop ?
>>
>> SWITCH or IF
>>
>>
>
> This is a little test I did. I get about 35ms for 'switch', about 20ms 
> for 'if'. As Jim said though, I suspect it'll vary according to the 
> kind and complexity of conditions you're testing.
>
>
> on mouseUp
>    put the millisecs into tStart
>    repeat 10000
>       put random(5) into tSw
>       switchTest tSw
>    end repeat
>    put the millisecs - tStart into t1
>
>    put the millisecs into tStart
>    repeat 10000
>       put random(5) into tIf
>       ifTest tIf
>    end repeat
>    put the millisecs - tStart into t2
>
>    put t1 && t2
> end mouseUp
>
> -----
>
> on switchTest pSw
>    switch pSw
>       case 1
>
>          break
>       case 2
>
>          break
>       case 3
>
>          break
>       case 4
>
>          break
>       default
>
>          break


[truncated by sender]