Niels Saurer

Some thoughts: I think there's nothing majorly wrong with the current way. If we wanted to change it, here's some suggestions: * `apply_` as prefix for the unit methods, and...

Has removing the layer of abstraction and using the raw bytes underneath directly already been discussed? AIUI, we use UnvalidatedX in cases where we just need something comparable to use...

I agree with @robertbastian that the "Or" is too inclusive here. But that does make me think, `StrAsBytes`/`BytesStr`? Otherwise maybe some non-negated antonyms of "trusted": `QuestionableStr`, `ShadyStr`?

My problem with these alternatives is that (imo) they either require more thinking than `UnvalidatedStr`, e.g., `SketchyStr`, or they are somewhat ambiguous, e.g., `MaybeStr` - to me this sounds like...

I agree with @Manishearth about the naming, `UnvalidatedStr` perfectly describes what the type is. If we have to avoid the negation, I like `LaterStr` the most out of all alternatives...

More tests have been added since that PR, in particular in the spec-draft-compliance-PR #3670. They cover a good range of use cases. What's still left is to decide whether we...

Hi @zelomel, thanks for the bug report! I've published a new release (v0.2.1), could you try that one? I didn't have time to test it yet. EDIT: You may additionally...

I assume this has to do with the new `sync15` protocol reMarkable is switching to - sadly I am still on the old protocol so it's difficult for me to...

I was interested in something similar, so I experimented a bit and ended up with `RVec` being refined over a predicate `p: (int, int) -> bool` that is defined as...

Thank you for the detailed reply, interesting thoughts! This seems like a challenging problem indeed. (As a side note, I'm really fascinated by this project, thanks for working on it!)...