My DB doesn't allow duplicates. But I need, rather than psedo-random numbers, to generate a 32-bit unique ID on Linux. I know someone has done this before. Does anyone know the link(s) of this software or of the similar kind?
Thanks
Jack

How unique does it have to be (i.e. unique within a single system or unique among a set of systems)?
The simplest method: associate a counter with your database and increment it whenever you read its value. If you modify this counter from several concurrent threads, you have to lock it on every modification.

Which database are you using? Most databases (RDBMS's like mySQL, pgSQL, Oracle and MS-SQL) have an "auto increment" attribute that will automatically generate a unique primary key value for you each time you insert a new record.

Otherwise, maybe the easiest technique is simply to store your "last index" somewhere (an external text file; maybe even a table in your database itself), and then increment "last index + 1" as part of your standard "insert".

'Hope that helps .. PSM

PS:
Please do consider building a small "project library" and please do consider the books I mentioned in reply to your post last night!

PPS:
A proprietary system I once worked on had a custom database called "MKF" ("Multi-Keyed Files"; it was designed and invented in the mid 80's, around the time relational databases were beginning to take off), and it had a table called "scalar_numbers". Which just stored the current value of things like "the current highest document ID". It served precisely the kind of purpose you're looking for...

Is a simple increment sufficient? auto increment.
Do you want the number to be evenly dispersed across the potential range? Hash table

Or something else? - please be more precise.

real time clock?