Jared Hoberock

It's true that a range-of-ranges formulation would be more general, but I think that this is a case where Worse is Better. For the purposes of discussion, let's assume that...

Sean has a function like this called `IntervalExpand` in MGPU. It works like a merge, and should be quite a bit faster than the implementation in our example code.

Variadic tuple is a huge, risky change. Is there any evidence that it is ready or that the compiler is ready for it? Here are some questions I'd like to...

If you would care to file a new issue for getting scons to work on Mac, we can discuss the problems you encountered with unit tests there. Besides the functional...

1. Yes, as discussed in #624, I do not think a new preprocessor switch is viable. 2. Yes, all code in Thrust which assumes there is a maximum tuple size...

I don't think we've discussed unit tests -- there will need to be new tests which test the functionality of tuples larger than 10 elements. These should go in the...

Let's keep the scope of this effort focused on variadic tuple for now in order to maximize the chance of its success.

Yes, I agree that integrating variadic tuple will require quite a bit of work. Nonetheless, I do not wish to condition this pull request on a separate tangentially related request....

That directory structure I suggested looks good to me. Please keep everything confined to this pull request. Thanks.

> Do you know of a guide for finding the most recent cuda toolkit on the system? `which nvcc`? Also [this advice](https://stackoverflow.com/questions/36279045/where-did-cuda-get-installed-in-my-computer#comment60198878_36279045) from Robert. Does the `--cuda-path` option put circle...