>I'm viewing this from the perspective of taking a set of data and
moving it from one environment to the next (ie, from dev to staging or
staging to production, all within the same database type).  In a
previous job we used integers as PK/FK and when we went to move data we
had a mess on our hands because there's no guarantee that the integer
will be unique across databases.  We had to build routines for jockeying
integers when promoting data.  No fun.

You also can't guarantee the receiving database will support identities
at all, or in the same manner.

>I like integers because they are easy to work with, smaller, and
faster.
I'm almost at the point of saying any record needs an identity integer
(for work within a system) and a GUID (for work across systems).  Does
this seem like a crazy approach?  I probably will not go this route
because of all the additional maintenance that goes with it.  But it's
an interesting thought.

If you are concerned about portability, I would use a GUID generated by
the application (CF or whatever).  Then, store the GUID in a char field.
GUIDs are _supposed_ to be unique across all systems.

M!ke
----------------------------------------------------------
You are subscribed to cfcdev. To unsubscribe, send an email
to [EMAIL PROTECTED] with the words 'unsubscribe cfcdev'
in the message of the email.

CFCDev is run by CFCZone (www.cfczone.org) and supported
by Mindtool, Corporation (www.mindtool.com).

An archive of the CFCDev list is available at 
[EMAIL PROTECTED]

Reply via email to