Marijn Schouten

Actually, is there really any reason for this complicated dance? Why not just match on "POST "? It seems that would probably do fewer bounds checks as well. Or if...

Right, well this patch just inlines what the compiler was asked to inline, possibly a slightly more efficient version, so there should be no performance regression. Is there anything you...

Alright, after some thought I agree with you that this is not very nice, since we are breaking out of the Bytes abstraction, so I am closing this in favor...

Right, there is only a single use as you noted, here: https://github.com/seanmonstar/httparse/blob/97c7e6e23672df1482930bdcd5ff1eae3dbecf16/src/lib.rs#L843-L867 and the computed pointer is at most one element past the end of the `bytes` slice, but that...

@kornelski I have my own reasons for wanting to disable autodetection of nightly sometimes, so I wrote a [small utility to override version-based detection](https://github.com/hkBst/rustc-with-version). Let me know if it is...

The core of the algorithm is: ```rust *hash = hash.rotate_left(ROTATE).bitxor(input).wrapping_mul(MUL); ``` so if hash starts out as 0 it will quickly become non-zero. So if the zero is problematic for...

Thanks, @lnicola, maybe next time I'll try it myself. There is one semantic change, that you may want to be aware of, which is that instead of only preserving the...

Done: https://github.com/cbreeden/fxhash/issues/20

Are deletions leaving unusable entries, that can only be reused by rehashing the entire table, the only reason that capacity can go down?

I think the issue is that the hashtable does not check whether the key is already present before deciding to grow. The following only does the duoble insert for the...