The end is exclusive for integer indexing but inclusive using dates. I guess if the data is irregular it's more weird to find the first data point after what you want to include, so we shouldn't change the current API. However it is still a little confusing so just wanted to note that here.
Comment From: wesm
At some point maybe an API method for non-inclusive indexing would be interesting
Comment From: wesm
This is subsumed by more general rules about integer vs label indexing