Rob Ede
Rob Ede
replacing the feature is a breaking change, can we adjust this PR so that it's additive instead?
I think you're right. My reaction was too cautious. I haven't been able to contrive a breakage. I'm going to push an update to 0.25 of hickory-resolver.
This field was made private for v4.0 but later it was realized this was the wrong trade-off. Making it public again is certainly planned for v5.0. This is a breaking...
The thing we were seeing in v3.x was users reporting confusion using this type without destructuring because the Deref impl allowed access to tuple elements like `path.1` but then `path.0.1`...
No rush. Like I say, this will need to wait for v5.0.
Can I ask what's your use case for wanting these impls?
See my comment in #3579
Our channel implementation is a lightly modified version of the one from `futures-channel`. I see this commit from a while back https://github.com/rust-lang/futures-rs/commit/ed3b268225c7c532a4b222adfe077c4e8ce0f24d which indicates there is a reason for using...
In the effort to promote sensible defaults, I feel that the positive impact of -Dwarnings far outweighs the negative given how easy it is to disable. CI is more often...
Lowercase header names are used as a normalized representation, since header names are case-insensitive in HTTP/1.x and _required_ to be lowercase in HTTP/2. Also see https://github.com/hyperium/http/issues/228#issuecomment-404996750