Closed daslu closed 4 years ago
I'm pretty sure they can. The error referenced in zulip discussion is a result of improper translation of logical columns
My conclusion is that transfer big amount of data is possible and works. What doesn't work is conversion of logical types to tech.ml.dataset
Next release of tech.ml.dataset will cover this issue
It's fixed already in t.m.ds
Rserve has a maximal socket buffer size, and that limits the amount of data we can pass through it.
Our solution can be to wrap the
r->clj
andclj->r
functions in a way that will transparently break the data into chunks before passing it over the socket.