>>To be honest, I think the ansi SQL should provide us with some mechanism to do more or less what we did in VFP for years. Not looking too strict on the fields missing an aggregation function, but by default take the last (non null) value as VFP did.
>
>I understand what you are saying, however doing that would render the ANSI SQL Standard syntax able to accept queries that have ill defined results according to the statement itself. Knowing the data doesn't cound when you are setting a standard. The non-aggregarate fields in one of those selects are not well defined in their return value so the standard disallows them.
>
>This problem is not really new to VFP9 it was a problem when trying to run queries that ran fine in VFP in SQL Server or Oracle where they violated the standard. Although I can understand what you are saying I don't agree that having VFP9 enforce the standard is a bad thing. Just have to get used to it and write queries that comply with it. Don't pull those extra fields in the original query rather get them after the first query with a seocnd one.
Actually, FirstVal() and LastVal() aggregate functions would give us the best of both worlds - they'd be fairly simple to implement, therefore fast, and would count as aggregate functions. Now if we only had a way to influence the standard :).