Once again I need to incorporate a mechanism which will issue a unique number as an object discriminator.
I'm sure we've all been there. - Incorporate a timestamp - quick and almost always unique, though this application will be busy, so the multiple 'threading' is attractive, Big problem is the length though, I only have 10 bytes.
Maybe pack the numbers into printable alpha numeric? use a radix of 36?
A record on a file (yawn) slow, need to Q for a lock, but a flexible way of doing it, and understandable by 'users'
a data area - its a file really, but maybe quicker to operate? same locking issues
A User space? - are they going to be quicker than data areas?
So - anyone any ideas? I can manage the bit about making it unique across systems by adding a boxID prefix (Darn - another byte lost!)
While I'd like to go with timestamp. I'm thinking I'm forced to use a sequential ascending number, so any helpful info / discussion about access times would be good.
July 25, 2011 8:53 AM
March 31, 2012 8:22 PM