depends".
1) Strongly typed datasets rock, they are faster than
untyped, use intellisense... but your reason for
wanting to use them is important. I use them every day
of my life, but I also don't use them every day of my
life and there are reasons for both.
2) How much of your process is dependent on reads vs.
updates/inserts? Datareaders are the way to go if you
aren't updating or your updates are limited. Remember,
you can always load a DR into a DT.
3) As far as having multiple Datasets. If the data is
the same and your question is for X tables, should I have
one or multiple Datasets, in general, One dataset is
preferable. However, you can use a Datarelation for
instance, from a Table in DSx and another in DSy just as
easily as you can from DSx.1 to DSx.3. One's a little
easier to code however. If the data that they hold is
different, in a Functional Dependency Sense.... then
seperate them. The way to think about a Dataset in most
instance is the same way you think of a Database. Does
it make logical sense to sepearte them? It doesn't cost
me anymore to have a Database with 300 tables or 2 with
150 tables, until I start trying to code them. Each has
it's mertis.
And, if from my experience and screw ups.... Don't
confuse DataTables with Datasets. You can relate a table
that isn't a member of a dataset to a table in another
dataset, just as easily as you can one that is in a
dataset.
Datasets aren't always necessary.
Also, if you have large query results and multiple users,
you WILL Kill your server and or network if you use
disconnected data. It's sometimes best to use a
Datareader and transfer the data into datatables.
I love ADO.NET, but it can be a great strain on
resources, and loads tend to grow which only exacerbates
the problem.
If you have specific examples, perhaps I can be of more
help.
Let me know,
Good Luck,
Bill
-----Original Message-----
I have a windows form which consumes quite a lot of data.
I am wondering what the best approach to managing the data is. I want touse strongly typed datasets, so should I :
Create one large dataset containing all the relevent table schemas andpopulate as required,
create individual datasets that encompass logical groups of tablesor does it not matter ?
Thanks
Andy
.