large data joins using collapse?

Hi, I just wanted to ask why you took *collapse* out of the 50Gb join benchmark, given that it had initially successfully completed it? *collapse* has a single-threaded implementation of joins which is very memory efficient. It should be able to complete the task. 