http://tools.ietf.org/html/rfc5646#section-3.1.7

'''
   In cases where the subtag or tag also has a
   'Deprecated' field, then the 'Preferred-Value' is RECOMMENDED as the
   best choice to represent the value of this record when selecting a
   language tag.
'''


Example - validation of the tag "no-nyn":
http://validator.nu/?doc=data%3Atext%2Fhtml%3Bbase64%2CPCFET0NUWVBFIGh0bWw%252BPGh0bWwgbGFuZz1uby1ueW4gPjxoZWFkPjx0aXRsZT5DaGluZXNlLCBNYW5kYXJpbjwvdGl0bGU%252BPC9oZWFkPjxib2R5PjwvYm9keT48L2h0bWw%252BDQo%253D&showsource=yes

Expected result: Recommendation to use "nn" instead.

Actual result:  As expected.


ISSUE: Inconistent treatment compared to the treatment of "no-hognorsk" (Bug 876)

* Currently "no-hognorsk"  triggers an error due to lack of use of "nn" as prefix, 
   while "no-nyn" triggers a warning because the preferred value is "nn". 

* However, in BCP47, the use of "nn" as prefix for "hognorsk" carries the same level of specification requirement as the  recommendation to use  "nn" instead of "no-nyn". That is: In both case, the choice of "nn" is RECOMMENDED, but not a MUST.

Whatever the outcome, the HTML5 validator should give the same treatment of *all* SHOULD level requirements in BCP47. Thus, if you insist on describing it as an error to use "nn-hognorsk" and not use "no-hognorsk", then - whenever there is an alternative preferred-value, you should also describe it as an error to  use grandfathered tags (such as "no-nyn" ) in those cases where the grandfathered value has a preferred-value field - in the case of "no-nyn" the "nn" subtag.

My preferred solution is that this bug remains as is, and that bug 876 is fixed to be in tune with this bug. (Thus that 'no-hognorsk' causes a warning and not an error.)

Alternatively, you could  - as suggested as an option in bug 876 - display an error message on the condition that you a) do the same thing in bug 876 and b) use correct error message language - SHOULD/RECOMMENDED.