|
dev
newsgroups
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Saving Dataset- Order of operationsHi
If you save a Dataset, which order must you do the operations in? I think its 1) Update 2) Delete 3) Add Or does the order matter? Vayse Its ok, I remembered it was in Sahils book.
Tis 1) Inserts 2) update 3) Delete Show quote "Vayse" <vayse@nospam.nospam> wrote in message news:%23DRg8JoUGHA.5288@TK2MSFTNGP14.phx.gbl... > Hi > If you save a Dataset, which order must you do the operations in? I think > its > 1) Update > 2) Delete > 3) Add > > Or does the order matter? > Vayse > Why is this neccessarily the right way?
I would argue Deletes need to take place before Inserts. For example, let's say the user deleted a row with primary key A. But then inserted it again. So the new row has a primary key A, and a bunch of other data that differs from the origiinal row which has been deleted. If you process the Inserts first, it will fail due to a primary key violation. If you process the Delete first, you will delete the old row (what the user did), and then you can Insert the new row - and everything will be fine. Show quote "Vayse" <vayse@nospam.nospam> wrote in message news:u84Y8hoUGHA.4900@TK2MSFTNGP12.phx.gbl... > Its ok, I remembered it was in Sahils book. > Tis > 1) Inserts > 2) update > 3) Delete > > > > "Vayse" <vayse@nospam.nospam> wrote in message > news:%23DRg8JoUGHA.5288@TK2MSFTNGP14.phx.gbl... >> Hi >> If you save a Dataset, which order must you do the operations in? I think >> its >> 1) Update >> 2) Delete >> 3) Add >> >> Or does the order matter? >> Vayse >> > > "Marina Levit [MVP]" <someone@nospam.com> wrote in message Because Sahil said so! ;) (p352 of Pro ADO 2)news:uhMegIpUGHA.5004@TK2MSFTNGP11.phx.gbl... > Why is this neccessarily the right way? Actually, you make a very good point about the primary keys. But would it also not depend on the way the key was set up? If it was an autonumber key, the insert would get the next number anyway. But in the case of the user entering the primary key, then I would need to do it the way you suggest. Better to do it that way in general I reckon. Thanks! Vayse If you're processing a dataset with related records, you'll want to process
deletes from the bottom up and then apply updates and inserts from the top down. For example, assume the following relations: Customers Contacts (of a customer) Phones (of a contact) Delete any phones, then any contacts, the any customers. Unless you cascade deletes and then you only delete the Customer record. Insert any Customers, any contacts, and then any phones. Updates technically can happen either with the deletes or the insert unless you alter primary keys in which case you either need to cascade udpates or actually delete from the bottom up and insert from the top down. |
|||||||||||||||||||||||
Other interesting topics