Paule,

I agree that count() functionality is universal and required in pretty much all non-trivial applications, and that therefore it was very surprising to me that the early Hibernate API provided support only reluctantly in patch form!

In the real world, every hibernate developer was working with patched jars and/or home grown "double criteria" hacks to accomplish one of the most common use cases!

An API should be designed for its commonest usage scenarios. All applications building interactive queries (i.e. probably any decent size webapp using hibernate) surely need pagination functionality. You rarely want to say "give me an unlimited number of results and send them all back to the client".

Luckily H3 provides the projections technique to accomplish this without patching. BTW the documentation didnt help me figure that out, I had to look at the source code.

As for using perl, well that's not such a bad idea ;) We've been looking at how we can speed up our development process. Writing Java web apps using standard "best practice" is incredibly inefficient. In the time it takes to build a simple CRUD app in Java you can write a whole portal in PHP or mason. However, tools like Eclipse and Hibernate are awesome and make us stick with Java, but just adapt our development models so we can work faster and smarter.

There's a reason Amazon use perl, Yahoo use PHP etc...but we're sticking with JSTL/freemarker for now.

Ben

Provide a documentation patch if you don't like the docs.

once you set projection to null you cant re-use projection with same criteria even with a detached criteria...any work around?

This thread is getting near to my post in this forum:

http://forum.hibernate.org/viewtopic.php?t=956564&highlight=pagination

I am looking for a way to use hibernate to get not only the pagination, but also the row count in a single statement. Taking the approaches in this thread would require you to do TWO distinct sql queries.

Using straight Oracle SQL there appears to be a way, but I don't know how to construct the Criteria object.

http://forums.oracle.com/forums/thread.jspa?threadID=510229

Anyone?

This approach doesn't work if the original criteria contained a projection, since you are overwriting it with null. I think you need to save the original projection that the criteria might contain AND the original ResultTransformer and restore them afterwards (as opposed to just setting it to null).

I describe my approach in the following thread: http://forum.hibernate.org/viewtopic.php?t=974802 (Pagination with Projection.rowCount() in a single call?).

(The fourth post from Octuber 31)

I am also implementing pagination with count by manipulating the projection for two different queries (page, count). I have set up a function to abstract pagination for me, to which I pass a Criteria object. The problem I am seeing is that the count query is significantly slowed down by the "order by" clause, which is unnecessary. I am getting queries that end up looking like this:

select count(this_.id) from employee order by hireDate desc

The order by clause in this case is unnecessary. I would like to run the count query second so that I can wipe out the order by, but it looks like all you can do is append to the CriteriaImpl's List of order by clauses.

Do you guys know any way around this other than modifying the CriteriaImpl class to have a removeOrder function?

Hi Paule,

I always wondered why is uniqueResult returning a Long and setFirstResult requiring an int ?


Kind Regards,