>Is it just me or is
No it is not only you...
I have often wondered why the syntax was not modeled after the select, ie:
INSERT INTO alias VALUES ;
( expression1 as field1 ;
, expression2 as field2 ;
, expression3 as field3)
if valuelist is smaller / not in order of actual fieldlist. Even "remodeling" existing inserts on altered tables would be easier to code IMHO. But I doubt fox heads will change SQL standards <g>.
If I program applications, I bite the bullet and nearly always use insert-sql, becuase in security/less chances for error vs. programmer convenience the convenience looses. YOur argument on buffering has merit, but who knows if there won't be a special mode with buffering disabled. When I am datamining/scripting large tables, on large insert lists I usually use replace for the reasons you mentioned, but that is not a multiuser environment and work is done directly on the tables and restarted in case of any error - definitely NOT the typical UC.
regards
thomas