Hello, I am thinking of creating a table with a field as such-

CompID:CompWebLink:True/False:DD/MM/YY:DD/MM/YY:True/False:0/60x200::

- short explanation of above array entries -
(CompID),(URL of Link),(URL still exists?),(Date last checked this DB),(Date link last worked), (is link on front page), (Banner link size)

Would i be better creating a new table for each CompID?

Absolutely.

You need to study up on database normalization. This is a set of rules concerning database table design which have been in existence since about 1972. A database is said to be normalized if it complies with these rules.

The very first rule essentially states that the data in a column be atomic, that is, comprised of simple, indivisible values. This form precludes multivalued attributes (groups) and composite data, such as the column you suggest.

Tables which violate first normal form can be very difficult to query against. For example, try to write a query against your proposed structure which lists all the links which worked last month...

Jeff Mason
Custom Apps, Inc.
[email protected]

Thankyou very much for your reply. I'm self taught on SQL and as such I've not had a basic understanding. You've pointed me in the right direction and now I feel I've got a grasp on at least one fundemental!

You made a good point on the difficulty of querying my column, thankyou for saving me wasting half a day creating a bad DB!

Mark

I've been looking at creating a Dynamic Table for each User:

SELECT @SQL = 'Create Table [dbo].' + @UserID + '('
SELECT @SQL = @SQL + '[UID] int IDENTITY (1, 1) PRIMARY KEY CLUSTERED, [ID] nvarchar(10),'
SELECT @SQL = @SQL + '[LinkExists] bit, [LastChecked] datetime, [LastWorked] datetime,'
SELECT @SQL = @SQL + '[FrontPage] bit, [BannerSize] nvarchar(20) )'

Exec (@SQL)

I don't think this is a good idea as all my queries would require Dynamic SQL to access a Database that matches a UserID. Also I've done a bit of reading on using Dynamic SQL with @TableName variable and it suggested better Database design is to use just one table and pull the rows with a query by a UserID column. Sounds like the best option all things considered.

You are much, much better off with a single table rather than an entire constellation of them.

Once again, consider the difficulty of querying. Analysis across the set of users will be next to impossible ("How many users had links which worked last month?").

Jeff Mason
[email protected]