What is the thinking for using separate function names (eg. json_extract for e functions vs json_nextract for n functions) as opposed to an extra argument "nullonerror" flag (eg. json_extract(j, nflag)? Is there a particular space/time trade off, or is the choice a result of simply adding a n function after the e function was written? So while the utility of e and n functions is pretty clear, the dual name nature is less so (at least to me). I am not suggesting one or the other, but simply curious as the thinking behind it.