[Hibernate] Roadmap - HQL

[Steve E. <ste...@jb...>]

Regarding the HQL translator related changes mentioned in the previous
email...

Mainly, this is involving a fairly large scale refactoring of the Antlr
grammars and the associated code.

The first aspect was previously discussed on this list, so I won't go
into all the gory details.  Basically, the current Antlr-based
translator translates an HQL query in 3 phases where each phase is
represented by one of the grammars.  As part of the refactoring, it was
decided to move to a 4 phase approach.  Essentially, the old second
phase will be split out into 2 separate phases (that's slightly
simplified, but the basic gist).

Another aspect was overall simplification of the AST trees.

This is work that has been on going, separated out on the HQL_ANTLR_2
branch in SVN.  The focus of the work on that branch thus far has been
the first and second phases.

This was all being done in an effort to add specific new capabilities to
HQL as well as make it easier moving forward to add even more new stuff.
For example, one of the new features (that is essentially done on that
branch) is the capability to perform "ad hoc" joining.  Another is UNION
and MINUS capabilities.

Another thing that was done was to change the manner in which "property
paths" are resolved.  The previous version delegated to the individual
nodes that made up the path after the entire path had been recognized.
That put a lot of responsibility on the nodes in terms of knowing where
they were located within the overall statement structure since that has
a bearing on how the path resolution needs to be handled.  I am not
completely satisfied with some of the details about how this is
currently being done on the branch, but it's pretty close to how I see
this being done in the eventual solution.  The one thing that is causing
serious issues with the current approach is index operators; they throw
a hugely gigantic monkey wrench into that processing currently :(