Sort/select processing when sorting by associated fields

[Boaz Bezborodko]

Kenneth Brody wrote:
> Quoting Boaz Bezborodko (Wed, 30 Jan 2008 12:55:36 -0500):
>
>>
>>> Date: Wed, 30 Jan 2008 12:36:50 -0500
>>> From: Kenneth Brody <kenbrody at bestweb.net>
> [...]
>>> You can verify this by creating a simple file.  For ease of setting
>>> this up, create a file with five instances of A1, and add 5 records
>>> filling in only 3 instances for each record.  Then, create an output
>>> sorted on A1.  Finally, create a sort/select processing:
>>>
>>>     mesgbox "Record " < @rn { "\n at AF = " < @af { "\nA1 = " & A1
>>>
>>> When you run the output, you will see that your processing is run
>>> 15 times -- one time for each non-blank instance of A1.
>>>
>> How does doing the Sort/Selection process on an indexed NON-associated
>> field affect this?  I'm asking since my approach is obviously messing
>> things up some since it only ever goes through the first record once
>> despite how many associated fields there are.
>
> Take my example, add an additional non-associated field.  Build an
> index on this field, and run your output with the appropriate "-i"
> flag.
>
>> BTW, I found a workaround that I am satisfied with for now.  Instead of
>> jumping to the first instance in the index of records I want to choose,
>> I pick a spot a little before it.  So if 'ba' holds the start date of
>> the selection set based on index 'O', the line looks like this.
>
> Unless, of course, you are looking for the very first date in the file.
>
> [...]
>
Would a solution to this whole issue (one that would include all 
instances) be to index on the date field (20 in my example) AND the 
associated field?

Boaz