Testing a data tier

[Tilted]

Evening chaps

I'm after a bit of advice, I've written a data tier using attributes,
reflection and inheritance, to test it I've written a simple order entry
program. Glad I did as it picked up many flaws, but as you may be aware as
soon as reflection is used things start slowing down, can anyone advise on a
decent test scenario.

I could create a table with 30 fields, create say 1000 records, create
another table with a 1:m relationship on the first with another 20 fields
and 5 records per record in the first table, and so on and so on.

Is this what you guys would do?

Chris.

[Kevin Aubuchon]

If it's order entry, I'm assuming an order will have one or more line items.
This lends itself to a parent-child table relationship as you mentioned.
Suppose the parent table is called Order, and the child table is LineItem.
For each order,
there will be one record in the Order table. That table may have a column
called order_id to identify it. Each line item on the order will have a
single record in the LineItem table. The LineItem table will probably have
a column called lineitem_id (uniquely identifies the line item), and another
column maybe called order_id. This order_id column will have the same value
as the Order record it belongs to.

Order id = 1001

LineItem id = 1 order_id=1001
LineItem id = 2 order_id=1001
LineItem id = 3 order_id=1001

This has been a very brief answer to your question which I think is really
asking a relational database theory.

kevin aubuchon
www.aubuchon-design.com
"Tilted" <pu***@gvygrq.pb.hx (ROT13)> wrote in message
news:%2****************@tk2msftngp13.phx.gbl...

Evening chaps

I'm after a bit of advice, I've written a data tier using attributes,
reflection and inheritance, to test it I've written a simple order entry
program. Glad I did as it picked up many flaws, but as you may be aware as soon as reflection is used things start slowing down, can anyone advise on a decent test scenario.

I could create a table with 30 fields, create say 1000 records, create
another table with a 1:m relationship on the first with another 20 fields
and 5 records per record in the first table, and so on and so on.

Is this what you guys would do?

Chris.

[Tilted]

Not really a question to do with relational databases, I've been using them
for years, hence the ability to develop this (apparantly) working tier.

The question was what is the best way to test it, should I just create
masses of data? Is this a true scenario? Is it better to just carry on
with the order entry type of thing or can anyone suggest a project I can
create which will thoroughly test its capabilties and speed.

Chris.

"Kevin Aubuchon" <ke************@charter.net> wrote in message
news:iw*******************@fe07.lga...

If it's order entry, I'm assuming an order will have one or more line
items.
This lends itself to a parent-child table relationship as you mentioned.
Suppose the parent table is called Order, and the child table is LineItem.
For each order,
there will be one record in the Order table. That table may have a column
called order_id to identify it. Each line item on the order will have a
single record in the LineItem table. The LineItem table will probably
have
a column called lineitem_id (uniquely identifies the line item), and
another
column maybe called order_id. This order_id column will have the same
value
as the Order record it belongs to.

Order id = 1001

LineItem id = 1 order_id=1001
LineItem id = 2 order_id=1001
LineItem id = 3 order_id=1001

This has been a very brief answer to your question which I think is really
asking a relational database theory.

kevin aubuchon
www.aubuchon-design.com
"Tilted" <pu***@gvygrq.pb.hx (ROT13)> wrote in message
news:%2****************@tk2msftngp13.phx.gbl...

Evening chaps

I'm after a bit of advice, I've written a data tier using attributes,
reflection and inheritance, to test it I've written a simple order entry
program. Glad I did as it picked up many flaws, but as you may be aware

as

soon as reflection is used things start slowing down, can anyone advise
on

a

decent test scenario.

I could create a table with 30 fields, create say 1000 records, create
another table with a 1:m relationship on the first with another 20 fields
and 5 records per record in the first table, and so on and so on.

Is this what you guys would do?

Chris.