On 16/09/2016 19:50, Rob Conde wrote:

The first case (int -> float) was fixed, but the second (float->int) wasn't.

Oh :(

I fear this may have to stay broken for this release - what I need to do is devise a test case that tests conversion of everything-to-everything - and that's going to be a lot of permutations!

On second thoughts, the second case (float -> int) is a conversion we have never supported (unsafe/lossy) - which is to say this is by design. That said, we have always supported conversion to native (built-in) integer types, so we're inconsistent there.

John.

Can you clarify your approach going forward? When this compiles (other versions of gcc, visual studio) it works as intended. It's not a huge deal for me whether this is fixed for 1.62, but I definitely don't want it to be expressly forbidden going forward.

Rob