I explicitly did not put it on crates.io and would strongly prefer it not
to be put on crates.io and that any instances of it be removed. It’s
obsolete and people should no longer be using it. It's already
Cargo-compatible, so the only reason for it to be on crates.io is so that
other crates.io packages can depend on it, something that people should not
be doing. They should probably use Hyper instead.

This library is obsolete. Please let it die.

I have to agree with you on that point. Every convenience addition to this library will extend its lifetime and make it more rooted in the ecosystem. I could agree to replace the dependencies to make the transition to something else a little smoother, but not replacing them may motivate people to finally abandon it. I do, on the other hand, think that suddenly rendering it unusable may be a bit harsh. I don't know how much it clashes at the moment. It may be acceptable for an obsolete library on its way to the graveyard.

I think the hardest-hit targets here would be nickel, since they build on rust-http. I managed to port Iron over to Hyper, but it took quite a bit of time and there was a lot of breakage that then coincided with the other change of moving to crates.io.

The change itself will not be perfectly smooth, but having two usable
libraries will at least let depending projects change when there's time.