rekordcrate
rekordcrate copied to clipboard
Error Handling
We should think about proper error handling facilities. Currently we are just passing along the errors that our parser libraries (nom
, binrw
) generate. IMO these are dependencies that should not appear in public interface, since that forces a hard transitive dependency for the library consumer, which in turn would make changing internals (restructuring parsing/serialization) a breaking change.
https://blog.yoshuawuyts.com/error-handling-survey/ https://nick.groenen.me/posts/rust-error-handling/