Closed DeliciousHair closed 1 year ago
Issue solved in https://github.com/JErnestoMtz/rapl/commit/09a0df5f37079e4e7d5cff020ac3ff569a669424#diff-f686acba59f5e44ba9e3177784a8c42d45806e266474c5540305fd33dad01f40.
Make sense to remove Tuples it just add noise. Now you can reshape with arrays up to length 20 i.e. Ndarras of 20 dimensions, that would probably take care off 99% of the use-cases. But for the other 1% it is also possible to do higher dimensions by first initializing your own Dim
object:
// for Dim<R> where R > 20
use typenum::U24;
let dim_24 = Dim::<U24>::new(&vec![2;24]).unwrap();
//2^24 = 16777216
let arr_dim24 = Ndarr::from(0..16777216).reshape(&dim_24).unwrap();
Using an
array
works great and is very flexible. Using atuple
on the other hand runs into obstacles if the dimension is greater than 4 at present. Given that using tuples requires additional implementations for each dimension, using tuples represents code bloat and a point of confusion as a user--which one should I use and why?My thinking is that removing tuples altogether makes thing easier for the user (only arrays work) as well as keeps the code bloat to a minimum.