Sebastian Berg
Sebastian Berg
This was brought up, and I am honestly wondering if it isn't better to just flip it. And if that means writing a short NEP because we skip deprecation... Then...
> Hard to get quantitative data, but the feeling is that it's very common to see At least for those specific users a warning is actually nicer, since they are...
Aha? I thought it sometimes fails due to funky lapack/blas implementations. And short of them having terrible bugs that we should maybe not work around, I would assume that is...
I forgot to point out that the important thing about byte swapping is a byte-swapped `-0.0` not a `0`, since it actually _is_ all zero.
And I forgot to say, thanks splitting it out is much easier considering how unwieldy especially the `.dtype` one is!
> This is not flagged by the linter. It is E251 (https://docs.astral.sh/ruff/rules/unexpected-spaces-around-keyword-parameter-equals/) should I add that to the configuration in a different PR? Runnnig it on the current codebase gives...
> PR created. The long lines is about 1500 changes which is maybe a bit too much? With ruff check --select E501 | grep "Line too long" you can see...
Sorry, this slipped, let's give it a shot, good time now for deprecations just after the release. Thanks @eendebakpt.
The 4 ULP SVML thing is partially a thing of the past unless I am confused: we did run into issues with sin/cos and eventually switched the highe rprecision version...
I don't mind if we try to use the SIMD fused ops also for scalars, but we don't get to a point where results between Python and numpy or comparing...