Size of array

[Mark Smith]

Yes, but you can make the keys sort of multi-dimensional: someArray 
[a,b,c], so for some purposes this works just as well.

In fact (note to shari), depending on your particular application,  
you may find it more efficient to have an array with 300 * 1000  
elements than to have an array with 300 elements of 1000 items each.  
Though rev's chunk expressions are extremely efficient, accessing the  
thousandth item in a list is generally going to be slower than  
accessing an array element, even in a big array.

Put this in a button script:

on mouseUp
    if the optionkey is down then
       makeArrayWithItems
    else
       makeArrayNoItems
    end if
end mouseUp


on makeArrayWithItems
    repeat 300
       add 1 to c
       repeat 1000
          put random(125676) & comma after sArray[c]
       end repeat
    end repeat

    put the millisecs into st
    repeat 10000
       put random(300) into c
       put 900 into d
       get item d of tArray[c]
    end repeat
    put the millisecs - st
end makeArrayWithItems

on makeArrayNoItems
    repeat 300
       add 1 to c
       put 0 into d
       repeat 1000
          add 1 to d
          put random(125676) into sArray[c,d]
       end repeat
    end repeat

    put the millisecs into st
    repeat 10000
       put random(300) into c
       put 900 into d
       get tArray[c,d]
    end repeat
    put the millisecs - st
end makeArrayNoItems

on my machine, though it seems to take a little longer to build the  
no-items array than the with-items array, access is nearly 4 times  
faster for the no-items array. The bigger the data, the more  
pronounced then effect (try making it 500 x 2000, or 1000 x 5000).

Of course it's unlikely that you'll be accessing the 900th+ item all  
the time, but it illustrates a worst case. Accessing the 1st item  
every time produces a tie, so on average, the no-items array is going  
to be faster.

Best,

Mark

On 18 Nov 2007, at 01:56, Petrides, M.D. Marian wrote:

> Speaking of arrays... Am I correct in understanding that Rev only  
> supports one dimensional arrays or am I misreading the dox?
>
> On Nov 17, 2007, at 7:54 PM, Mark Smith wrote:
>