I have the map shown below in one of my entities:

I really need the db that NHibernate creates from this mapping to have a clustered unique index created that includes the columns evaluationId and varName. This is simple to do in SQL Enterprise Manager after the db is created, but can NHibernate's mapping file express this so it's done automatically?

In general, I don't know how to get a unique constraint that includes more than one column. I know about <property name="..." unique="true"/>, but if I want the constraint to span more than one column, what can I do? I haven't been able to find this topic in the documentation.

Thanks in advance!

Try this:


i.e. use <column> tag with unique-key attribute.

That did the trick. Thanks! And thank you so much for working so hard for us! NHibernate rocks.

Is a clustered index too SQL Server-specific to hope for a mapping notation to indicate it within NHibernate? That is the only thing lacking in the mapping you suggested.[/i]

I just noticed that the code sergey suggested earlier in this thread is not quite what I was looking for. That hbm mapping generates a unique constraint across two columns, but not an index across those columns. I really need the columns to be indexed. How can the mapping file be changed for that?

I've looked in NH and Hibernate's documentation without finding any pages on how to use the <key> or <index> tags. I'm probably just looking in the wrong place.

Thanks.

Ok, so I was able to find some documentation for it.

http://nhibernate.sourceforge.net/h2.0.3-docs/reference/html/toolsetguide.html

The following will accomplish my goal: change the unique-key attributes to index attributes. This actually had the effect of turning that unique constraint into a clustered, primary key index. Excellent.