Heikki Linnakangas escribió:
> I do understand the point, though - it's much easier to edit and debug
> long statements when the value is close to the column name. I find that
> the INSERT .. SELECT makes that a lot nicer:
>
> INSERT INTO t
> (col1,col2,col3,col4,col5,col6,col7,col8,col9,col10,col11,col12,col13,col14,col15)
> SELECT 'val1' AS col1,
> 'val2' AS col2,
This example lists the columns twice, which is lame (you have to keep
both in sync) -- and if you take the first list out it works, but the
values can end up in the wrong places if they are not in the same order
as the columns in the table.
--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.