I tried to follow all the good job the 4 Hackathon's guys are doing.
I saw you were trying to integrate the atPoint() method written some
weeks ago by Cameron. As I remember this was implemented to help the
user query features to retrieve attributes. When there are several
vector layers, events cannot be detected trough the svg tag that is on
top, and then it's not possible to detect clicks or mouseovers on
features drawn in a bottom layer (ie. not on top).
A thought we had at this time but without testing it was to have a
single svg (or vml) for all the vector layers. Not just the simple div
tag for each OpenLayers layers.
Each vector layer would be differenciated using group tags.
This may break the OpenLayers architecture but this would probably
help for events handling.
A great advantage of this solution (if it works) will also rise up
when integrating the editing tools and the snapping modes. If user
needs snapping on features from an other vector layer we currently
have to simulate it (PointArraySnapping). This means to loop every
points of every features.