Hibernate version:2.1.7 or 2.1.6 (tried both)

Mapping documents:Middlegen 2.1

Name and version of the database you are using:Firebird 1.5.1

Hi, I have a table called 'Collections' and another called 'BorrowerTypes'. There is an 'overrides' table (called Overrides_Collectns_Brwrtypes) that associates the two and is a many-many. i.e There can be an override between one collection and several borrower types. (Of course these tables have been renamed by middlegen.)

I want to list the pk of all Collection rows, so in HQL I do this:

which gives me 51 records, which is correct.

I also want to indicate in the results whether (for each row) there is 0 or more 'overrides' records. This could either be a count of 0-n or a List of 0-n records, it doesn't matter.

So I modified the HQL:


However I get the error
Examining the SQL in the hibern8ide console, I see

The column c.overridesCollectnsBrwrtypes is a hibernate representation of my table; it seems that either hibern8ide or hibernate has not generated the correct SQL code.

Can anyone tell if the query should be done differently?
Could there be some sort of mapping error (in one of my mapping files)?

Regardless, hibernate should not be submitting this malformed SQL to the database - it is clearly using a hibernate-generated field name whose name differs from those in the database. It also hasn't resolved the relationship that exists between the Collections and Overrides_Collectns_Brwrtypes tables (i.e some sort of join).

TIA
PS. Clutching at straws, this also fails (same type of error):

First: Hibern8IDE just uses Hibernate to execute HQL so it doesn't do any sql-generation of it's own.

Secondly: You are using a aggregate function in the select clause which when applied to a collection should be done (as written in the docs):



Read: http://www.hibernate.org/hib_docs/refer ... ggregation

;)

Good to hear, I wasn't sure, but was hopeful that the work was passed on to hibernate ;-)


Yes, I've read the docs, and that's what I was originally trying. But I was getting errors like 'elements is not a property of xxx' (sorry, cannot remember the exact message, but can follow up tomorrow at the office).

I discussed this problem a fair bit on IRC, tried many things and the final suggestion was to post a message here :)
Like I say, even if the HQL is malformed, I'm suprised that hibernate could generate SQL which contains hibernate-generated property names.

Cheers

The parser in Hibernate 2.x doesn't have the capability to detect the correct syntax. The LL(K) lexer/parser/parser combo in Hibernate 3.x is an excellent new feature done by Joshua Davis. We'll have both systems in Hibernate 3.x and you can switch back if there are any issues (we have thousands of regression tests, so this will hopefully not be neccessary).

Not that big a surprise since the logic is that if hql doesn't recognize something as a keyword or hql-identifier it will assume it is some sql function/identifier which can be quite usefull ;)

Uhm, yes thats by design. I hadn't read the original posting.