Hi Victor, I thank you for your efforts here, especially your addition of DirEntry.inode() and your work on the tests.

However, I'm a bit frustrated that you just re-implemented the whole thing without discussion: I've been behind scandir and written the first couple of implementations, and I asked if you could review my code, but instead of reviewing it or interacting with my fairly clear desire for the all-C version, you simply turn up and say "I've rewritten it, can you please review?"

You did express concern off list about an all-C vs part-Python implementation, but you said "it's still unclear to me which implementation should be added to Python" and "I don't feel able today to take the right decision. We may use python-dev to discuss that". But I didn't see that discussion at all, and I had expressed fairly clearly (both here and off list), with benchmarks, why I thought it should be the all-C version.

So it was a bit of a let down for a first-time Python contributor. Even a simple question would have been helpful here: "Ben, I really think this much C code for a first version is bug-prone; what do you think about me re-implementing it and comparing?"

TLDR: I'd like to see os.scandir() in Python even if it means the slower implementation, but some discussion would have been nice. :-)

So if it's not too late, I'll continue that discussion in my next message.