>>Granted, so to ensure that a duplicate does not exist, you can add a machine_id to the string, then search the table to verify that the key has not been used.
>
>Searching a table only discovers data that has been saved to the network, not data still in buffers on individual machines, and the machine_id adds a long string to the ID number that is difficult to read and report.
>
>Why go through all this effort when you can use a lookup table? Just lock the appropriate record, increment the value and unlock the record for someone else to use. Unique, you can use VERY large numbers (I think it's 2 billion) if you use integer-type, and easy to handle.
>
>Barbara
It really depends on how the application is going to be designed to be used. I am programming all of my keys to be able to be used as distributed databases (remote and/or offline views), where access to a reference key table may not be available. I suppose that eventually, I may start using the windows "CSLID" function to generate the unique keys, if I run into the chance to create a globally distributed program (Please excuse me while I wipe off my chin!)
//:^)