Does the DataSet have a future?

[Greg Gates]

I have recently finished reading 'Hibernate In Action' by Christian Bauer
and Gavin King. I'm quite impressed with the functionality that is
available in Hibernate. I am now questioning if the DataSet has a future.

While the DataSet certainly contains a good deal of very useful
functionality, it appear that a Hibernate solution offers more flexibility
and, given good tools, a similar ease of use. In addition, the DataSet is a
very "Microsoft-centric" approach. I wonder if a DataSet style approach is
transferable to the Java world. I get the sense from following the Java and
Object newsgroups that a table-oriented architecture is generally not
considered to be the best approach as it is not as flexible and is more
difficult to implement complex business rules.

So what to do? NHibernate is still in beta. There are no out-of-the-box
tools for object relational mapping in .NET. There are a variety of third
party ORM tools for .NET, but they have widely differing approaches and
none appear to have the momemtum of Hibernate.

Is a true ORM solution superior to the DataSet, and will this cause the
DataSet to become legacy once ObjectSpaces is available?
It appears the Hibernate offers the functionality that ObjectSpaces will
offer, but I can use Hibernate today!

For these reasons,I am seriously considering using Eclipse,Hibernate and
SWT/JFace for a windows client desktop application. What do you think? Is
this going too far? Am I throwing the baby out with the bathwater?


thanks, Greg

 

[Cor Ligthert]

Greg,

Looks for me the same as those guys who wanted to stick to the punching
cards and described then why the magnetic tape had no future. They were
right the magnetic tape is not so much used anymore.

Just my thought,

Cor

 

[David Browne]

I have recently finished reading 'Hibernate In Action' by Christian Bauer
and Gavin King. I'm quite impressed with the functionality that is
available in Hibernate. I am now questioning if the DataSet has a future.

While the DataSet certainly contains a good deal of very useful
functionality, it appear that a Hibernate solution offers more flexibility
and, given good tools, a similar ease of use. In addition, the DataSet is
a
very "Microsoft-centric" approach. I wonder if a DataSet style approach is
transferable to the Java world. I get the sense from following the Java
and
Object newsgroups that a table-oriented architecture is generally not
considered to be the best approach as it is not as flexible and is more
difficult to implement complex business rules.

Look at JDBC Rowsets. New and very similar toe DatSets. I think Java is
just trying to catch up with ADO.NET.

So what to do? NHibernate is still in beta. There are no out-of-the-box
tools for object relational mapping in .NET. There are a variety of third
party ORM tools for .NET, but they have widely differing approaches and
none appear to have the momemtum of Hibernate.

Is a true ORM solution superior to the DataSet, and will this cause the
DataSet to become legacy once ObjectSpaces is available?

And exactly when to you think that will be?

It appears the Hibernate offers the functionality that ObjectSpaces will
offer, but I can use Hibernate today!

For these reasons,I am seriously considering using Eclipse,Hibernate and
SWT/JFace for a windows client desktop application. What do you think? Is
this going too far? Am I throwing the baby out with the bathwater?

Definitely a bad idea. DataSets are much better than ORM solutions for
windows client desktop apps. For that kind of app you don't want to hide
the tabular nature of the data. Fetch DataSets, bind them to UI widgits,
let the user fiddle with them, and flush them back to the database. You
will be able to this in a fraction of the code it would take with
Eclipse\SWT\Hibernate.

ORM tools aim to make relational data "look like" objects. Even if an ORM
tool is 100% sucessfull and introduces 0% overhead, you're still treating
living a lie. Tables are not objects, and vice versa, and pretending
otherwise will always cause problems.

Now depending on how data-intensive your application is, this may be a
trade-off you can live with.

David

 

[Val Mazur \(MVP\)]

I believe Java will not be able to catch .NET. Like for me it is too late
for them to catch it. It reminds me same situation with Sybase SQL Server.
They had pretty big market, but did nothing to make their product better. As
a result Microsoft punished them with their version of SQL Server, which now
has more market share.

 

[Sylvain Lafontaine]

As always when someone speaks this kind of language, the real question is
not to know if DataSet or Hibernate have a future but to know if they have a
present. I may safely assume that DataSets and Hibernate will be dead 100
years from now but - quite probably - so will I be.

An ORM model can be a good thing when you are dealing with something like a
trafic cars simulation; as each car can be seen as an object, each object
(car) behind different in many ways from other similar objects (model, year,
speed, consumption, etc.).

However, when you are dealing with things like sport statistics, shopping
carts, merchandises, etc.; dealing with these like if each value was an
object is just plainly stupid.

S. L.