Andrew Gallant

This feels like a good "first issue" to me. In particular, routines could be added that are implemented by first converting the input to a `&str` and then reusing std's...

@alexjago Thanks! Nice work. Some responses: * It should definitely take `AsRef`, not `&BStr`. `&[u8]` is what all of the APIs in `bstr` are defined on. * I'm not sure...

Requiring Rust 1.55 is fine. I'm going to bump the MSRV of `bstr` up higher. See the 1.0 issue.

You're saying an assert is an alloc? No, it isn't. Let's stick with how std does things I think wherever we can. The radix is almost always going to be...

Well, the assert might alloc when the assert fails. Sure. But that's okay.

> do the BStr and BString types pull their own weight? They feel a bit redundant to me -- I essentially only ever use the former and even then, only...

> For reference, not using &BStr in gitoxide would be a great degradation of usability of parsed git objects which can't be forced into UTF-8 encoding. Could you say more...

@Byron Ah wow, that is great feedback. I agree that using `BStr`/`BString` in those APIs is probably the right thing, because otherwise the derived `Debug` representation of those types is...

Ah perfect. My time is super limited and devoting pretty much all of what little free time I have to `regex`. But I'll do my best to take a detour...

Folks, I've updated the top comment of this issue to include a list of breaking changes (all of which are minor). My plan is to push out the 1.0 release...