Greg Stark <stark@mit.edu> writes:
> On Fri, Aug 12, 2016 at 7:40 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> pointing out that "SELECT 42 ISNULL" is now ambiguous. Since ISNULL
>> is nonstandard, maybe dropping support for it would be feasible.
> Isn't ISNULL one of the lexical tricks we used to effectively give
> bison two token lookahead?
No, it's a SQL-exposed feature, though I think it's just there for
compatibility with some random other DBMS. See also NOTNULL.
regards, tom lane