Supposedly valid `Language` instances may have invalid tags

[dscorbett]

Language.get normalizes case and enforces syntactic validity whereas Language.make doesn’t. This causes some discrepancies. I presume that a user is not supposed to have to worry about case or syntactic validity once they have a valid (in the sense of is_valid) Language object.

If Language is valid, its tag should presumably be valid. This is not always true.

>>> lang = Language.make(language='Latn')
>>> lang.is_valid()
True
>>> tag_is_valid(lang.to_tag())
False
>>> 
>>> lang = Language.make(script='Qaaa..Qabx')
>>> lang.is_valid()
True
>>> tag_is_valid(lang.to_tag())
False
>>> 
>>> lang = Language.make(extensions='t-fr')
>>> lang.is_valid()
True
>>> tag_is_valid(lang.to_tag())
False
>>> 
>>> lang = Language.make(extensions=['a'])
>>> lang.is_valid()
True
>>> tag_is_valid(lang.to_tag())
False
>>> 
>>> lang = Language.make(language='x-') 
>>> lang.is_valid()
True
>>> tag_is_valid(lang.to_tag())
False

Round-tripping a valid Language through its tag should presumably return an equivalent Language. This is not always true.

>>> lang = Language.make(language='FR')
>>> lang.is_valid()
True
>>> lang == Language.get(lang.to_tag(), normalize=False)
False

Alternatively, maybe it is the user’s responsibility to normalize case and check for syntactic validity before calling Language.make. I don’t think the documentation actually says that though.