I (and likely one or two others) will take this under advisement. I agree that it is not right, in a purist's sense. I had also spent time worrying over that wee conundrum, but ultimately decided that its conceptually clean resolution should wait for another day (or year.) Coming from an engineering background, I respect the adage, "the perfect is the enemy of the good." It is taken to heart by engineers who earn their keep. This is not to say there will be no resolution, which would probably be as you suggest. But I hesitate because, knowing Richard and Dan, (who wrote most and much of parse.y respectively), I am inclined to err assuming they know what they're doing unless that is clearly not the case for some little fact. The question I must ask, (and I intend no snidery with it), is: What is the likelihood that a SQL writer, with low interest in language-lawyering, will trip over this when the syntax railroad chart immediately precedes the operator table?