We got a bug report from Coverity:

*** CID 1411801:  Incorrect expression  (MISSING_COMMA)
/Parser/tokenizer.c: 111 in ()
105         "OP",
106         "AWAIT",
107         "ASYNC",
108         "<ERRORTOKEN>",
109         "COMMENT",
110         "NL",
>>>     CID 1411801:  Incorrect expression  (MISSING_COMMA)
>>>     In the initialization of "_PyParser_TokenNames", a suspicious concatenated string ""ENCODING<N_TOKENS>"" is produced.
111         "ENCODING"
112         "<N_TOKENS>"
113     };
114
115
116     /* Create and initialize a new tok_state structure */

I missed this typo :-p