GeezerVison

$ ./python -m timeit -n 1000000 -s "from pathlib import PurePath" "PurePath('a', 'b')"
1000000 loops, best of 5: 636 nsec per loop  # before regression (baseline)
1000000 loops, best of 5: 827 nsec per loop  # after regression (30% slower)
1000000 loops, best of 5: 662 nsec per loop  # after this fix (4% slower)

The last few % are hard to recoup unfortunately

Skipping news because this only regressed yesterday.

When you're done making the requested changes, leave the comment: I have made the requested changes; please review again.

I have made the requested changes; please review again

Thanks Alex

First alternative: it's been a while since I've dealt with this, but I think not calling super() from __init__() can snooker users of multiple inheritance in certain cases. Maybe? My editor warns me about it but I don't recall specifics.

Second alternative: would make the initialiser signature of PurePathBase and PathBase different from PurePath and Path, which I think would be hard to explain to users.

I think I was wrong about the first alternative: it's fine to omit the super() call. Will adjust the patch.

Yeah, if you're looking to do cooperative multiple inheritance, it's always advisable to always call super().__init__() when overriding __init__ in a subclass.

Outside of that situation, though, I don't think there's any reason to say you have to always call super().__init__() in a subclass -- there are often good reasons to avoid doing so. Linters often warn about it because a common mistake is to accidentally forget to include the super().__init__() call in the subclass -- but we know what we're doing here, so it's all good :)

The cooperative multiple inheritance concern doesn't apply here either, I don't think, since PurePath itself only has one base -- PurePathBase. But it's also been a while since I've tried to do cooperative multiple inheritance with anything; I might have to rewatch https://www.youtube.com/watch?v=EiOglTERPEo&list=PLd9YubueMBAAHenyO2yyf4bg27uwzx-Fc&index=11 to brush up :)

Second alternative: would make the initialiser signature of PurePathBase and PathBase different from PurePath and Path, which I think would be hard to explain to users.

Yeah, good point!

Thanks for the reviews, both!