On Thu, May 7, 2015 at 8:29 AM, Tim Dudgeon <tdudgeon.ml@gmail.com> wrote: That's not going to work. I want the row, I just don't want the values that are null.Only thing that comes to mind:1. Use the conversion function to get the json structure with nulls.2. Use an explode function to convert the json into a table structure with (key, value) columns.3. Filter that table where value is not null.4. Convert the remaining entries into arrays5. Pass the two arrays back into the json_object(keys text[], values text[])You could dynamically build up a literal string array but the syntax challenges scare me:json_object('{' ||CASE WHEN col1 IS NULL THEN '' ELSE '"col1",' || val1 || '"' END ||CASE WHEN col2 IS NULL THEN '' ELSE '"col2",' || val2 || '"' END ||'}'::text[])David J.
That's not going to work. I want the row, I just don't want the values that are null.
pgsql-sql by date:
Соглашаюсь с условиями обработки персональных данных