Phil Miller - NOAA
Phil Miller - NOAA
@crtrott You said in today's meeting "we figured [this] out". Please document the resolution, and remove the 'Blocks Promotion' label.
Daniel's PR fixed it for 3.7, and Nathan cherry picked that to develop. This is fully resolved.
4.0 released with this.
Removing InDevelop so that we don't think this is fixed and close-able at the next release.
Per Daniel, further review and integration is waiting on performance regression testing.
It looks like GTest does have some accommodation for this issue: https://github.com/google/googletest/blob/main/CMakeLists.txt#L32
I'll post a PR, I guess
Sadly, I'm still in an environment dealing with a lot of projects that do vendor it. I would hope the maintainers consider this worth accepting a fix for.
I think the ideal outcome of this would be to fold the scaling factors from right to left, with types as if the fold were initialized by `element_type{1.0}` -- i.e....
Conversely, how cautious are users expected to be in not inadvertently **widening** their calculations? If one had `mdspan m`, and said `scaled(2.0, m)`, rather than `scaled(2.0f, m)`, that's going to...