So far, we have 3 issues in this thread:
- OP's grouping in Uru
- OP's grouping and alscript
- My issue with duplicated groups
Grouping in UruI performed a test, and PyPRP does recognize the member objects of a group. PyPRP basically treats them like normal objects, and pretends the grouping does not exist. They all still export, and act like normal objects in Uru (without any grouping).
For those individuals who really want the grouped objects to work like a group, in Uru itself, parenting is the best alternative. Maybe it would be possible then to set up something in PyPRP where when a group is exported, a new empty is created with the name of the group, and the member objects are parented to the empty. Although a user doing this manually would probably be the best way of doing business.
Grouping and AlscriptI don't see how this would be possible until we have better tools for alscript handling in Blender. Right now all we have is a single large text file. In the future (probably with 2.5 and some of the coming changes for PyPRP) perhaps there will be some kind of handling system that works through a GUI that allows you set options for objects as individuals and as batches (and that will be handy for things like visregions). Then it would be possible to have a single script and be able to apply it to a number of different objects (1 palm script for 30 palm trees).
Duplicated GroupsAs a user you can add each of your objects to a group (and an object can be in multiple groups). You can then effectively create a clone of the group using either a Group Duplicator, or Dupligroups (dupligroup uses an empty and has that nice ability to re-assign what group it will clone).
And this can work between two files, meaning you can reference a group in blend file A, and the entire group, and all it's numerous associated objects will now be cloned in blend file B, yet easily movable (unlike a linked library) yet not editable (like a proxy object). And in fact, this is the only way you can reference a group from another blend file.
The major benefits of this being that if you are working on a large scene, you don't have to join all you objects into 1 mesh (making editing difficult), or otherwise linking a very large list of individual objects.
As I said before, PyPRP does recognize the objects in a normal group. I did some more tests and Dupligroup and Duplicators cloning a local group (no linking from another file) and they seemed to be ignored by PyPRP and are never exported to Uru. Naturally linking them isn't going to make a difference. PyPRP will ignore any dupligroup or duplicator, file linking aside.
It would be nice if somehow PyPRP could actually take a duplicator, grab it's references, and export them. Basically treat a cloned group as a normal group. So when it all ends up in Uru, the components of the cloned group are just normal objects.
I don't how exactly you would code this, especially with 2.5 not so far away. I'm not sure how groups work in 2.5, or if it's pointless to implement anything like this until later.
Here is a page containing information about Blender's Object Groups.