00:00 <nisanharamati> ?

00:00 <nemosupremo> I changed the CONCURRENCY var to 13, but left it 31 in the ab.to_state_partition function. I thought that was the reaosn why

00:00 <nemosupremo> (line 43 in the gist I sent)

00:00 <nisanharamati> oh

00:00 <nemosupremo> but when fixing it I still got a segfault

00:09 <nemosupremo> I have to go, I'll check the logs if you message on here. if you get a chance to check it out

00:12 <nisanharamati> I can reproduce the segfault

00:12 <nisanharamati> I need to go too, but I can investigate a bit more later or tomorrow morning

00:40 <nisanharamati> Nemo is out, so I commented in their gist, but to those following or interested: the segfault was happening because the return type of their `join` computation did not match that of a stateful computation. It was doing `return "xyz"`, which returns a list and works for `computation_multi`, but stateful computations require a second argument denoting whether or not to save a state update to the log, so the

00:40 <nisanharamati> correct return type would be `return ("xyz", False)` or `return (list("xyz"), False)` (in this case, where state isn't persisted). It would be `return ("xyz", True)` where state should be persisted for recovery later.

00:51 <nemosupremo> Right! I forgot that detail. You need the Boolean if you want to persist state. I guess I wouldn’t have hit this issue if I was using Go. Thanks for helping me out even if the issue was minor

04:38 <nemosupremo> Can I only define one serialize method in my application (the global one)?

05:33 <slfritchie> nemosupremo: correct

23:52 <Sundist> help