Robert Bastian
Robert Bastian
I feel like `unsafe` should be reserved for memory unsafety. Using it as a general might-violate-some–invariant marker will lead to more code being marked unsafe than necessary. Crates sometimes for...
I guess for now we flatten the enum like `runtime-manual-retain` and `runtime-strip`?
It doesn't, because then someone will set `base_language_handling` for hybrid or preresolved and that doesn't make sense.
Using `core::fmt::Error` means code like `formatted_date_time.to_string()` panics, so I don't think we should ever use it. We should GIGO, in debug mode we debug-assert, but in release mode we return...
We should use a different garbage string for each code location. `(icu4x error)` is not actionable.
The GIGO doesn't have to happen at the top level that you linked. `write_pattern` should be infallible, and then `write_field` would do garbage out instead of returning errors: https://github.com/unicode-org/icu4x/blob/main/components/datetime/src/format/datetime.rs#L239. This...
What are your criteria for "user friendly"? `{?}` certainly *looks nicer*, but this is an error, it shouldn't look nice. We also want to communicate which names are not loaded;...
The use case for this message is `DateTimeNames::with_pattern_unchecked`. The `unchecked` part clearly communicates to the developer that they have to be careful with this and verify the output before any...
> For numeric fields if the FixedDecimalFormatter is not loaded, load the FixedDecimalFormatter in no-data mode, but still indicate the error using brackets Why not use `Writeable for u8`? Also...