define .slice and dataframe methods

[brandly]

here we have an implementation of .slice, a method that takes a cursor_id and a range of time, and it returns all matching events.

we also have a dataframe method that maps each span over .slice.

TODO:

• .slice makes multiple HTTP requests if the requested span has a large number of events in it. currently, it just concatenates the list of events from each request into one big list. however, the python lib does some restructuring of the data, giving you events on a per-stream basis.

each individual response has a payload like

{ "streams": {
  "stream_id": { /* details */ },
  "another_id": {},
},
"events": [ /* lots of events */ ]}

there's an entry in streams for every stream involved in the given query. each stream ID maps to an object containing some details about the stream. i have access to those details, but since R parses the JSON into a list, i can't seem to find the stream IDs in the data structure.

• dataframe currently returns a list. i need to figure out how to construct a proper dataframe, but .slice should be improved first
• dataframe makes an HTTP request per span via .slice. in python, we do this concurrently via a ThreadPool. i'm under the impression that R is single-threaded, so i'm not sure what to do here. @apclypsr do i have options? it's currently quite slow.

[apclypsr]

You can use this package for multithreading, but need to be careful to define the number of cores used not as the number you have, but as number you have minus one. Otherwise R will take up all your CPU and crash your os

http://stat.ethz.ch/R-manual/R-devel/library/parallel/doc/parallel.pdf

[brandly]

i'm gonna merge this, since there's a solid number of improvements here. i'll open an issue regarding the return values of .slice and dataframe

thanks @apclypsr for pointing me to parallel. it's working on my end, and i'm under the impression that this implementation will work on Windows as well, although that hasn't been tested yet.

at this point, the API endpoints are mostly wired up. i'm shifting my focus towards query construction and figuring out what a decent DSL looks like in R.