query DataSet

  • Thread starter Thread starter DG
  • Start date Start date
D

DG

I have a copy of DB in my dataset (just rows that i need ofcourse).
I'd like to create a view. (connectiong tables via Foreign Keys, and getting
columns that i want)
Is there a way to query dataset, or should i go to database for that.
(i was hoping to save myself trip to DB, cause all datata needed is in
DataSet)

Thanx.
 
If you are looking for a view in the sense of combining the columns of two
different datatables, then there's not a simple object to handle it per se.
The DataView object is based on one table at a time so it can't be used. But
you can either loop through your existing rows in your dataset and build a
new datatable, or query your view directly from the db, which will give you
a local datatable that has the same schema and data as your db's view.

Is this what you were asking?
 
i had doubts what to do, to loop through rows, to go to DB again, or is
there a better way.
I must say in my case verdict goes to another trip tp DB, cause i have 3
tables and looping join would bee too expensive.

Thanx.
 
DG,
I must say in my case verdict goes to another trip tp DB, cause i have 3
tables and looping join would bee too expensive.
Click to expand...
I am always curious what expensive means in this, the framework is full of
loops behind the scene, so it cannot be the processing time?

Cor
 
DG,

You might be interested in the assembly I've been working on at
http://www.queryadataset.com. Besides INNER JOINS, it lets you perform
complex SQL SELECT statements including UNION, OUTER JOINS, GROUP BY,
HAVING, ORDER BY, sub-queries, etc against the tables in a dataset. The
result of a query is a non-updatable DataView.

Adrian Moore
http://www.queryadataset.com
 
DG said:
i had doubts what to do, to loop through rows, to go to DB again, or is
there a better way.
I must say in my case verdict goes to another trip tp DB, cause i have 3
tables and looping join would bee too expensive.
Click to expand...
DG- there's obviously going to be quite a few factors that influence this,
but a round trip across the network to a database, to pull data that will be
redundant since you're joining on the db - If performance is your main
priority, I'm not sure hitting the db is *necessarily* the way to go. And
if you loop, there's less possibility of General Network errors/
Connection/Command Timeouts etc.

Definitely not trying to be argumentative on this end - just figured I'd
point it out to be safe ;-)
 
Back
Top