add optional precompute parameter to take a list of discrete impedance values or range of values about pandana

[fscottfoti]

Hmm, to be clear, precompute doesn't precompute the aggregations, it computes and saves the nodes that are available within a distance. So if you were doing an aggregation within 3000 meters you would need all the nodes within 3000 meters. I don't think you're saying you want to do an aggregation with a hole in the middle, like a donut right? At any rate, there wouldn't be any way to save time - you'd still have to traverse all the nodes in the graph on your way out, and you'd just optimize the nodes you save, which would save memory. I'm not sure there's a way to speed up what you're asking for - doing regional queries at the pedestrian scale does indeed take a lot of time. Let me know if I'm misinterpreting.

[sablanchard]

@fscottfoti ah yes that is right all the nodes < the value still need to be traversed so it seems that a speed up cannot be achieved. Figuring out a way to pickle the pre-computed network seems to be the best option then. Will close the feature request unless there are any other ideas on this topic.