@mattip Can we add here the cleanups suggested by @eli-schwartz that I made in 1b47584 ?

Can we add here the cleanups

Yes. I thought I did. Adding.

@stefanv I prefer the bash syntax with the quotes [[ "$NAME" == "name" ]] rather than without [[ $NAME == name ]], if you don't have a strong preference.

@stefanv I prefer the bash syntax with the quotes [[ "$NAME" == "name" ]] rather than without [[ $NAME == name ]], if you don't have a strong preference.

I don't have any strong preference myself. It was pointed out to me that this is unnecessary, but it does provide consistency across the script, and doesn't cause harm.

I don't have any strong preference myself. It was pointed out to me that this is unnecessary, but it does provide consistency across the script, and doesn't cause harm.

I do have a preference (skip the quotes) but I originally didn't mention it because, well, I'm not a numpy developer so I will ordinarily never look at, edit, or execute this script. :P

Which is why I took care to point out in the other PR that "to quote or not to quote inside these special brackets" is a stylistic thing and has an effect on the reader's comprehension (to wit: some people feel the quotes help them understand how bash tokenizes it, so they find the quotes helpful)...

... and crucially, that the quotes themselves aren't the cause or solution of the build error.

Since the people who are actually affected by the style used here, either don't care or want the quotes -- it absolutely makes sense to include the quotes in a STY: commit. (But it is STY: and not BUG:)

self-merging to move this forward. We can keep iterating on this if there are still problems

@mattip sorry, been busy this week.

Did you see if a manual push uploaded a wheel?

I manually started the CI job and the wheels ended up at https://anaconda.org/scipy-wheels-nightly/numpy/files. ✔️