I agree with Brian:  if you are moving large amounts of data to/from  
disk and they are arranged in a predictable way, READ FILE# and WRITE  
FILE# are your friends.   I have found it possible to read a single- 
precision floating point array containing 48 million elements  from  
disk into memory (or vice versa) in a few seconds.

-- Mark Chappell

On Nov 6, 2008, at 2:19 PM, Brian Stevens wrote:

>
> On Nov 6, 2008, at 9:47 AM, Peter wrote:
> <snip>
>> Or is there a more modern way, this just seems clunky
>
> That code is just inefficient and a solution has nothing to do with  
> "modern" code. It will do approximately 3,800 Writes when it can be  
> done in 1 Write.
>
> The method is to pass the address of the array and do a Write for  
> the entire length of the array. The example below ( hacked together  
> quickly, so review carefully ) is a single dimension array but multi- 
> dimension is just a matter of calculating the size of the array ( or  
> hard-coding ) and using it. This uses an FSSpec but that is not the  
> example's focus.
>
> '------------
> begin record Stuff
> dim as str31    size
> dim as long     length
> end record
>
> begin globals
>
> dim as Stuff     StockR(10)
> dim as FSSpec    fs
> end globals
>
> //Clear array and load with test data
> BlockZero(@StockR( 0 ),sizeof( Stuff ) * 11 )
> StockR.size(3) = "large"
> StockR.length(3) = 12345
>
> // Write to TestFilePB
> fn FSMakeFSSpec( system( _aplVRefNum ), system( _aplParID ),  
> "TestFilePB", @fs )
> open "O",#1,@fs
> Write File #1,@StockR(0), sizeof ( Stuff ) * 11 // write out entire  
> array in one Write File
> Close #1
>
> // Read TestFilePB
> open "I",#1,@fs
> Read File #1,@StockR(0), sizeof ( Stuff ) * 11
> Close #1
>
> print StockR.size(3),StockR.length(3)// prove it is there
> print StockR.size(0),StockR.length(0)// prove these are zeroed
>
>
> do
>
> handleevents
> until gFBquit
>
> '-------------
>
>
> Brian S.
>
> --
> To unsubscribe, send ANY message to: futurebasic-unsubscribe@...
>