Nathan Goldbaum

> If not, then doing anything more than a cursorary one-off local: I agree with this if the library doesn't rely on global state and there's no danger of someone...

I wonder if it makes sense to also move the floating point status error handling code into its own header-only library as well. That would avoid duplicating it in `float16.h`....

> One naive question I have is how changing the shape attribute is more thread-unsafe than mutating the array data itself. It's also about static type safety. @jorenham might be...

Sounds like a good idea! I'm traveling this week so if you or someone else would lime to edit the original issue please feel free, otherwise I'll get to it...

@qulogic pointed me at matplotlib, which already does this: https://github.com/matplotlib/matplotlib/blob/3323161b83c56f90cf63d2b30c24a65b8b12c953/lib/matplotlib/_api/__init__.py#L405-L410

@liam-devoe have you ever seen anything like this? Somehow in hypothesis internals, pure-python expressions aren't evaluating to the correct result.

My worry about keeping all the pypy-specific code in-tree is that people will cargo cult it and it will likely be broken (obviously or subtly) without any testing. It's also...

> perhaps you would be interested in making a TSan version based on what I have here? I suppose to test that this is understandable and maintainable. Sure, is there...

> For SciPy, we should add some docs, yes, and it would probably be sufficient for a README here to link to those docs? Sounds reasonable to me. Thanks so...

> Seems like we are happy punting Windows down the line Yeah that's fine I've never tried to do this exercise with sanitizers on Windows and am not aware of...