> From: Gavin Brook [mailto:[EMAIL PROTECTED] > Personally I use an individual primary key myself. In the > past I have seen oracle databases with two fields as the > primary key, but never a whole record. The only justification > I can see for having the whole record as the primary key is > to save space. Adding an extra column to store a primary key > requires the space to store it and the storage for the > sequence. On modern servers storage is not usually a problem, > particularly with such a small field. On older mainframes, it was.
I can not speak to AS400 as that was the technology in text books 10 years prior my college experience (yes I'm a youngin'). Anyway, any space you "might" save will be lost when you do updates or selects in which you want a specific record assuming there is a comparable stored procedure concept on AS400. Consider the extra lines needed to determine which record to update or select: UPDATE .... SET .... WHERE a=1, b=2, c=3, e=4, f=5 etc etc etc. SELECT .... FROM .. WHERE a=1, b=2, c=3, e=4, f=5 etc etc etc. Compared to UPDATE .... SET .... WHERE a=1 SELECT .... FROM .. WHERE a=1 That's not to say that there is never a need for compound keys comprised of 2 or 3 columns such as those found in join tables, etc... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Special thanks to the CF Community Suite Gold Sponsor - CFHosting.net http://www.cfhosting.net Message: http://www.houseoffusion.com/lists.cfm/link=i:4:183715 Archives: http://www.houseoffusion.com/cf_lists/threads.cfm/4 Subscription: http://www.houseoffusion.com/lists.cfm/link=s:4 Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=11502.10531.4 Donations & Support: http://www.houseoffusion.com/tiny.cfm/54
