A Lexeme page currently has two places to enter a language code: the language code associated with the lemma and the language code associated with the representation. As a user I expect them to accept the same input. Currently the input for the language code of the lemma is correct and the input for the language code of the representation should be adjusted.
{F16745033}
Leszek dug up how the validation is currently done:
* the entered lang code is checked using validator provided by LexemeValidatorFactory::getLanguageCodeValidator (which to be less technically direct means it validates that the language code is the language code recognized as valid by "core" wikibase)
* in case the user input lang code is not valid that way, the extra check happens
* if the input contains a part '-x-' in it, the part before '-x-' is taken, and validated the same way as above. In other words. Assuming "de" is a valid wikibase lang code, and "foo" is not, for lemma and representation language the following stuff is considered valid: de, de-x-Q666, de-x-yolo. And the following are to be considered invalid: foo, foo-x-Q666, foo-x-yolo, de-yolo, de-Q666