Master detail relationships

2005-09-21 10:53:12 PM
First of all I leave this message yesterday morning at the multi tier group but I guess iīve chosen the wrong group.
Two years ago I was so sick of so many problems I found With master detail
relationships(phantom records),the record changed by another user soap opera and so on.
I decided to create my own "master-detail schema" ,basically I used a clientdataset and a
Dataset pair for every detail ,I closed and opened this detail dataset everytime it was
But of course this strategy have a lot of disadvantages,the big one is that it have to make
one call for evey detail dataset.
Today this application still runs and is very fast,but it runs locally and fortuanetelly
I wonīt have to make it a three tier one in a near future.
This was good for cases where I had more than 4 details for one master,So a detail is opened only
only when it was needed(Many times the user donīt want to see or change every details).
How does the datasetprovider works when you configure it to retrieve details on demand.
It retrieves all the details of the master or only those that are neeeded?
They send just one package with the data from the master and
the details togheter?
How many calls it makes?
How many of you had problems is the past,and how many of you are having problems now with master detail relationships.
How does other Languages/IDES treat master detail relationships?
I think Nesteddatasets to be a joke,I never was able to make them ,and donīt know if
theyīre a good idea(I mean moving to many things that are not needed ).I might be wrong
and Would like to know about cases of success and cases of failure,specially in three tier applications .
Iīm using Delphi 7 and betteradodataset.