In certain situations, you've already done some validation
on a type. For example, you may have checked that a path
is an absolute path, or that a string is valid UTF-8.
However, since there is no standard library support for
representing such types, maybe you continue passing that
value around as a string or a []byte. This has a few downsides:
- Reduced dev velocity: It is not clear to the reader which operations are safe to perform, and editing the code becomes harder as it is unclear which invariants must be upheld.
- Poor debugging experience: Failures may appear much further downstream in case validation was not done on all code paths.
- Performance cost: Out of defensiveness, different parts of the code may perform the same validation redundantly.
Following the Parse, Don't Validate mantra, it is useful to have a dedicated type to represent the validation, so that it is clearer to the reader which operations are safe to perform and which operations require extra care.
This library exposes a few types such as UTF8String, UTF8Bytes,
AbsolutePath and RelativePath to represent common validations.