Hi, Cliff, and thanks for offering your help.
All possible keys are in the main table.
The remaining tables are organized the same way but are incomplete; they
lack some matching rows and or they lack some data. I thought it would
be better to combine the columns into 1 table because all the data is
available for search purposes without having to complicate a high volume
application with many shifting variables in the query. To arrive at this
stage I have already optimized nearly 400,00 lines of tab- delimited
text in 15 files (tables) to get to the present stage of 4 tables of
under 4,000 rows. Once established, the data will remain virtually
constant for a period of one year between scheduled updates.
So the answer to your question is 'yes', but that is only a 'side
benefit' as I see it. All of this comes from years of work with C-ISAM
which is why I am so unsure of working with unions and joins.
George
On Sat, 2008-10-25 at 23:59 -0500, Cliff Nieuwenhuis wrote:
> G. J. Walsh wrote:
>
> > 1 of the 4 tables has the complete range of keys, in this case about
> > 1,000. The other 3 tables have data representing most but not all of
> > those keys, and in different sets. I want to end up with a new
> > 'combined' table which will allow me to immediately 'see' missing data
> > from the 3 smaller tables and take the necessary steps to 'fill in the
> > blanks'. I realize I would of course have to create that table without
> > constrictions other than the primary key.
>
> Do you mean you want to find keys that exist in the 'main' table but
> have no match in the other three tables?
>
> --
> Cliff Nieuwenhuis
>