Manish Goregaokar

https://github.com/rust-diplomat/diplomat/issues/103

Let's do it

I would caution against looking to the stdlib for consistent naming, the stdlib fossilized names before conventions existed, often. But in this case I think either is fine; I don't...

I don't have a strong opinion here, but I have been bitten by the lack of all timelengths in the past.

> My priority then was avoiding the disaster of naming the zoneless type `DateTime`. I was willing to trade extra typing for `PlainDate` and `PlainTime` in return for avoiding decades...

At the moment? It would have to go through the FFI layer. In the future we can make this configurable on Diplomat so that it automatically does this for C++....

I don't see how this issue precludes any of that. For APIs that support different kinds of inputs we can have different methods. We can even add the unvalidated one...

Okay, so here's a plan, which I discussed a bit with @sffc The general idea is that for 1.0 we will expose `std::string_view` APIs but not assume valid UTF8 (except...

> Why do you say `&[u16]` needs to be validated? What UTF-16-taking APIs do we have that don't already deal with unpaired surrogates? Both the collator and the normalizer deal...

With #2534 and #2542, the 1.0 plan has been fully implemented. This is now a 1.x issue.