Re: [HACKERS] PATCH: recursive json_populate_record()
От | Michael Paquier |
---|---|
Тема | Re: [HACKERS] PATCH: recursive json_populate_record() |
Дата | |
Msg-id | CAB7nPqR_su1w5yC2KMSuH=XmmY8JHL+aDz-sJzc3Ln-juOAgtQ@mail.gmail.com обсуждение исходный текст |
Ответ на | [HACKERS] PATCH: recursive json_populate_record() (Nikita Glukhov <n.gluhov@postgrespro.ru>) |
Ответы |
Re: [HACKERS] PATCH: recursive json_populate_record()
|
Список | pgsql-hackers |
On Tue, Dec 13, 2016 at 9:38 AM, Nikita Glukhov <n.gluhov@postgrespro.ru> wrote: > It also fixes the following errors/inconsistencies caused by lost quoting of > string json values: > > [master]=# select * from json_to_record('{"js": "a"}') as rec(js json); > ERROR: invalid input syntax for type json > DETAIL: Token "a" is invalid. > CONTEXT: JSON data, line 1: a The docs mention that this is based on a best effort now. It may be interesting to improve that. > [master]=# select * from json_to_record('{"js": "true"}') as rec(js json); > js > ------ > true > > [patched]=# select * from json_to_record('{"js": "a"}') as rec(js json); > js > ----- > "a" That's indeed valid JSON. > [patched]=# select * from json_to_record('{"js": "true"}') as rec(js json); > js > -------- > "true" And so is that. > The second patch adds assignment of correct ndims to array fields of RECORD > function result types. > Without this patch, attndims in tuple descriptors of RECORD types is always > 0 and the corresponding assertion fails in the next test: > > [patched]=# select json_to_record('{"a": [1, 2, 3]}') as rec(a int[]); > > > Should I submit these patches to commitfest? It seems to me that it would be a good idea to add them. -- Michael
В списке pgsql-hackers по дате отправления: