I know that this is probably a topic that has been hashed and rehashed, but searching these forums doesn't seem to turn anything up about "is null".

I'm wondering why queries still need to use "is null" instead of allowing null comparison with "=" which is converted behind the scenes into an "is null" query that's sent to the actual db.

If I do:

or if I do the following with :name bound to null:

then why can't that be translated in the background to be


I understand tri-state logic and the reasons for having it in databases (specifically so that table joins work the way they do). I don't understand why we have to maintain that separation in the syntax of the query language. It just creates hassle and I have never understood what we gain by this.

As a result of these problems, I have in the past used a specific value to indicate null rather than using null itself so that I do not need to create one version of the query for each combination of is null. That's a completely ridiculous thing to have to do.

In other situations I've build the query with a ternary operator for each nullable field like


Is there a reason for continuing to require "is null"? It just seems like madness to me.

I though about that one night with Gavin, and yes there is a reason to keep HQL as a ternary-logic based query language. There were some queries you could not achieve otherwise.

I can't remember what it was but by thinking hard, one can reach the same conclusion.

What sort of queries can't you do? I've asked others this question before and no one has ever actually given me an example of a query that requires it.

Considering that:
....ANYTHING = null
resolves to false for all values of ANYTHING, what possible queries could use that? If you need an always false condition then you can just do "1=0" or some other always false condition.

it does not resolve to false, it resolves to null, that's a fundamental difference. It mean the negation of anything = null is also null, not true.
As I said, I can't remember the reasoning, hence the queries, but by digging around what I said you should find the family of queries you can't reach.

Ok, so if there are queries where it's useful then why not flip things around? Why not have "=null" behave like "is null" and "is null" behave like "=null"? If you wanted to avoid confusing people, you could have used "is unknown" or some other keyword.

That way, the infinitesimal corner cases where this sort of thing is needed would be covered while the far more common cases would not require the torturous query jimjamming that I described above.

I know that nothing is going to change now. I love hibernate and have used it for years (and am now starting to use ejb3). It just seems so sad that this one problem wasn't dealt with when you had the chance to change things.

The amount of Java developers is much smaller than the amount of SQL developers.