Hi all,

I have an entity who's behaviour is determined by a 'category'. However, within a category we have many (>200) different sub-categories (which do not affect behaviour).

We let a user create an entity by first letting him choose the category and then choosing the sub-category. The sub-category is persisted and category is not.
This is an existing enterprise system, and I can not just alter the tables to allow this. :(

The discriminator should be the category, not the sub-category. Not in the least because we do not want to create hundreds of subclasses.

Therefor, is it possible to use a sub-query to determine the category instead of making it a column in the table? We have translation tables that will determine the category for every sub-category value.

I don't understand what you want to do; what will this "sub-query" do exactly ?
I assume that you want to say: "I have an graph of entities; and the type of an instance is determined by a 'category'" !?!

Can you give a pratical example (with a minimal graph and some usage code) ?

Hi,
I think I know what TheShark wants - he has a 'discriminator' column containing values 1,2,3,4.....,200 but 1,2,3.....,50 means Class1 , 50..130 means Class2 and the rest Class3 (for example).
In my opionion it is not possible to create "calculated" discriminator because of one reason - discriminator MUST be read-write data, which cannot be in the case of calculated data. I'm afraid that if it is absolutly impossible to alter the data structure, you will have to resign from polymorphic approach here.

michal is exactly right. That is what I want, sorry if my explanation was not clear. :oops:

However, in my opinion, a calculated discriminator is possibly since it only needs to be calculated when read.. It does not need to be written. When the class is written (for instance class 1, but value 20), it also writes that value (20). So we only need to be able to translate the value 20 to Class1 at instantiation time. (using michals explanation here).

Although we have altered the data structure in the particular case, I'm not sure if I'm going to be able to do the same in other cases, so i'm still very interested in a reply! :lol:

If ( 1, 2, 3, ..., 50 means Class1 ) then what is the difference between them (1, 2, 3, ..., 50) ?
And when do you use that difference?

There is no difference in behaviour. However, the difference is important for the application.
For instance, you can have only one of each at a given period in time.
Which means there can possibly be 50 Class1's per client.

Are you sure that you are not misusing the table-per-class-hierarchy mapping strategy ?

Yep, in the current situation, the chosen strategy seems correct. But, it doesn't matter, we have added the 'category' column so all is great now.