Closed cwoodall closed 6 years ago
Thanks for the issue and pull request! It’d be nice to confirm that it’s not a temporary bug but rather a deliberate change to the type inference. If you don’t have any more insight into this issue, I propose waiting a bit. It might go away tomorrow or the day after.
A little more detail about this. I think the issue is that for Complex::i()
the generic type <T>
of Complex<T>
is not explicitly defined. I found that the following also worked (though I like the original solution better). The change might have been in the num_complex
library rather than in rust.
let sign= if inverse { Complex::<T>::i() } else { -Complex::<T>::i() };
Rust Nightly 1.23 breaks dft builds. To remedy this I explicitly types a few variables which seems to have eliminated the issue. My guess is something changed about the way types are derived which made something unspecified. I believe this fix should be backwards compatible too.
This addresses issue #10
Tests still run