RF - replace _filter_peaks with unique_vertices

[MrBago]

This unique_vertices will be used by the HemiSphere and for fiber tracking.

[MrBago]

Maybe just filter_vertices? Any other ideas for the name of a function which takes a set of vertices and returns a subset of those that are separated by at least some angle theta?

[Garyfallidis]

filter_vertices could be using many different filters. The filter you are using is angular distance. So, the complete name and awesome name is angular_distance_filter !! :-)

[MrBago]

How about remove_similar_vertices? Any takers?

[Garyfallidis]

Yeah sure. Make the change and then I will happily do the merge.

[arokem]

@Garyfallidis This breaks the test suite (with several import errors). Please don't merge things in w/o making sure that the tests don't break. I think Stefan is working on fixing this right now. I will update you on this.

[Garyfallidis]

Cool thx.

[Garyfallidis]

I would love to hear what you guys think is the best merging strategy.

[arokem]

First try rebasing on top of the current master. Resolve any conflicts that occur when you do that. Then run the test suite. If the tests don't pass, go back to the pull request and tell everyone about it. Sort of like what I did here: #33

Iterate until all the tests pass and everyone is happy. Then, merge the feature branch into master (never the other way around!) and push master back to nipy/dipy

[MrBago]

Generally I agree with you Ariel, but in this case we're rewriting several core modules so I've been trying to get them pulled one at a time so I can get feedback. I've found that very large pull requests do not get as much feedback. I'm working on updating the tests as we speak.