Christopher J. 'CJ' Wright

For me this issue is a bit broader, debugging complex pipelines can be rough. It might be nice to have ways to make it less rough. Another approach could be...

```python import time from streams import Stream from operator import add import matplotlib.pyplot as plt def star(f): """Take tuple and unpack it into args""" @ft.wraps(f) def wraps(args): return f(*args) return...

@mrocklin is there a way to have the map function delay asynchronously? It seems that the map (and it's timer sink) are actually executed before the b timer sink.

@martindurant I think it would be good to have coverage of the tests. We can `pragma: no_cover` things that we won't cover.

I've been trying to exclude the test chunk from the library chunk. I think the coverage is low because we don't run many of the dataframe tests. The coverage of...

1. Many of the tests are cudf tests, which we can't really run on travis. We can leave them in, but we'll need to list them as no cover. 1....

Do those cover the purpose of this PR? This PR was supposed to add an additional check to see the coverage of the test code itself. That way we could...

No problem, I should really merge master into this

Maybe we have an accumulate which takes in two streams, the first has the data and the second exists just to send a reset message (or even a new start...

Just as a heads up, I would not use `next` as an arg name since it shadows a python builtin.