Currently, both matching and validation are done on the basis of regular expressions. The advantage to this is that matching and validation are just data and can be edited without changes to the codebase. The disadvantage is that regular expressions are limited in the validation they can do. The most prominent example is that ISBNs can't be checked for validity, only for being the right "shape".

One way to improve on this would be to allow either providing a regex or the name of a built-in function defined in code, such that e.g. ISBN-13 could be specified to validate using an `isbn13` built-in, written in JavaScript, which does full checksum validation of the ISBN. For the majority of identifiers, nothing would change.