I wonder where the selection in Open GL is actually done. On the server
or client side? I'm asking because I don't want to use a big selection
buffer and evaluate it when all the objects have been draw. I'd rather
like to enter selection mode, draw one of the objects, switch back to
render mode, test if there was a hit, and do the same with all other
objects. If the selection test is done locally, all these mode switches
shouldn't decrease the performance too much. (Please, correct me if I'm
wrong.) However if it's a remote server that does the calculations,
there must be an unacceptable round-trip for each test.
I cannot check this out, because I'm using Mesa where everything is done
locally, but I wan't to be compatible with real client-server
implementions. Can anybody shed some light on this?