On 12/9/15, jeremy at copiousfreetime.org <jeremy at copiousfreetime.org> wrote: > So far it looks as if the way to distinguish between a json key existing > with a > value of null vs. not existing is to use `json_type` > > sqlite> select json_type('{"a":2,"c":[4,5,{"f":7}]}', '$.x') IS NULL; > 1 > sqlite> select json_type('{"x": null, "a":2,"c":[4,5,{"f":7}]}', '$.x'); > null > > Is this correct?
Yes. > > Also are there optimizations in place so that a column that is a json string > is only > parsed once if it is involved in json1 functions? For example: > No. The JSON parsing turned out to be so fast that such optimizations didn't seem worth the effort. Of course, things might change in the future. -- D. Richard Hipp drh at sqlite.org