Manish Goregaokar
Manish Goregaokar
Seems fine. Should we just return a Cow there too?
> 1. Is it a breaking change to add a concrete impl of a to_string function on a type that already implements Display? > 1. Is it a breaking change...
If it returns `Cow` i'd perhaps name it something appropriately cowish
The basic issue was that the Cargo resolver does not do well with `~` dependencies, and that is extremely unlikely to change. What ends up happening is that Cargo can't...
That crate only gets the soft guarantee (and no semver relaxation): for me the line there is "still compiles and mostly works" (as in, it's an okay intermediate, temporary state...
That's not the proposal. The proposal is, two parts: - icu_locid and icu_provider drop ~. Nobody else, except potentially some icu_util crate if we ever make that - icu_locid, icu_provider,...
With the soft guarantee, whenever we change internal doc-hidden/unstable APIs (say, in, icu_locid_transform) we ensure that downstream crates on older versions still _compile_ with the new API, perhaps by keeping...
For sealing traits I would suggest that we just use the typical `: Sealed` pattern. We don't need any documentation string because it is impossible to implement outside of our...
> Question: when you say "use the typical `: Sealed` pattern", what exactly is `Sealed`? Is it a per-trait type that is defined alongside the trait being sealed? Do we...