The only reason I didn't go with the HQL_ANTLR_2 branch was that it
wasn't "mavenized" at least as far as I could tell. That was the
easiest way I could get a working project up and running within my IDE
(mvn idea:idea). Could we fork off a branch from the mavenized trunk
to do this work? Then we could work on it together if you've got some
time to set up the 4-phase approach.
On 9/5/07, Steve Ebersole <steve(a)hibernate.org> wrote:
On Wednesday 05 September 2007 01:48:31 pm James Carman wrote:
> I would like to get this thing working (the join ... on ... syntax
> stuff). I have begun work on it already (got a build working, started
> digging into ANTLR grammar, etc.), but I would appreciate any guidance
> that you folks could provide. I started by adding a couple of test
> cases and trying to figure out where in the heck this stuff goes
> haywire! Right now, I'm looking into the SqlGenerator class as it
> seems that's where the meat of this is going to be. I'm not an ANTLR
> expert, so that's probably been taking me the most time to wrap my
> head around. Again, any help you guys can provide would be greatly
> appreciated, both by me and my client, since they're footing the bill
> for this development effort (we need it on our project).
> hibernate-dev mailing list
The very first thing I would do is to split the roles currently filled by
FromElement into two separate node types (one for the join and one for the
persister reference) because currently this is all munged together. It'll
make this easier to work with.
The big difficulty is somehow tracking the property references as they are
being resolved so that you know what is the right side and left side of the
This whole problem is easier to solve when we move from 3 phases to 4 phases,
which is the reason this was put on the back burner in the first place.
There is a branch where i started prototyping some of this work:
might want to take a look there for some ideas.
hibernate-dev mailing list