Re: File Join question

Rob,

Once again, you read between the lines very well.  Multiple programs/sessions 
will be accessing the index.

I am unwilling to create permanent "cheater" indexes because the base PFs can 
be VERY large (30M-50M records).  As indicated in my earlier response, my 
current approach is to create a join logical where the non-primary fields are 
after all the primary file fields, compiled with LVLCHK(*NO).  My current plan 
will be for these join files to be deleted on a nightly basis, with the option 
of making them permanent.

Regards,
Michael

> Now, if there is just no way to build the right indexes, and SQL keeps 
> doing a copy then I can see your pain in the performance hit.  Especially 
> since I get the impression that multiple programs need to access this same 
> path in a days time.
>