A traditional goal in database design is to place data in "Third Normal
Form" which means in essence that each data element is only stored in
one place. Acesses join the rows to deliver data.
A normalized database does not hold redundant data and changing the
value of one element changes its value everywhere.
You are proposing a database which is more like :First Normal Form" and
is deprecated.
T&B wrote:
Hi John,
Thanks for your reply.
You would have a better database if you normalize it and not do what
you propose.
Meaning what, exactly?
Obviously, I've only given rough indications as to my data source, such
as:
the huge Products table (which is actually a UNION ALL of various
supplier catalogs)
without detailing where all the data in there comes from, since that
would distract too much from my question. But suffice to say that
"normalizing" beyond the current structure is not straight forward.
In any case, at least in my ignorance of what you propose, it's beside
the point.
In essence, I'm asking: is it possible to update multiple columns in a
row, where all those values come from a single related row, without
SQLite having to find (eg via SELECT) that related row multiple times?
Or, put another way, I want to get SQLite to:
1. Locate the related row.
2. Grab the desired columns from that row, putting each in the related
row.
I hope that clarifies.
Thanks,
Tom
-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------
-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------
