dependent. So on anything other than a completely heterogenous > cluster, you will run into random segfaults, illegal instruction > errors, etc... >
(I think you meant homogeneous :-) I also work on a heterogeneous cluster. I try to get around this right now by logging into the same nodes (I don't want to give up my fast startup time yet!). But it also haven't done much dustributed work yet, so I'll have to delete sys.so at some point too. Cheers, Kevin
