On 04/15/2013 11:16 AM, hubert depesz lubaczewski wrote:
> In current 9.3, I see:
>
> $ select p.proname, p.provolatile from pg_proc p join pg_namespace n on p.pronamespace = n.oid where n.nspname =
'pg_catalog'and p.proname ~ 'json';
> proname | provolatile
> ---------------------------+-------------
> json_in | s
> json_out | i
> json_recv | s
> json_send | s
> array_to_json | s
> array_to_json | s
> row_to_json | s
> row_to_json | s
> json_agg_transfn | i
> json_agg_finalfn | i
> json_agg | i
> to_json | s
> json_object_field | s
> json_object_field_text | s
> json_array_element | s
> json_array_element_text | s
> json_extract_path | s
> json_extract_path_op | s
> json_extract_path_text | s
> json_extract_path_text_op | s
> json_array_elements | s
> json_array_length | s
> json_object_keys | s
> json_each | s
> json_each_text | s
> json_populate_record | s
> json_populate_recordset | s
> (27 rows)
>
> Is there any particular reason extract functions
> (object_field/array_element/...) can't be immutable?
>
> I can't readily imagine a situation where output of these functions would
> change for different queries.
>
Me either. It's an oversight, really. Unless there is any objection I'll
change them toot sweet. What about the existing (as of 9.2) functions?
cheers
andrew