Re: SQL/JSON: JSON_TABLE - Mailing list pgsql-hackers

From Andrew Dunstan
Subject Re: SQL/JSON: JSON_TABLE
Date
Msg-id cb6ce329-2899-9a2c-80d9-62284f2e570f@dunslane.net
Whole thread Raw
In response to Re: SQL/JSON: JSON_TABLE  (Andrew Dunstan <andrew@dunslane.net>)
Responses Re: SQL/JSON: JSON_TABLE  (Greg Stark <stark@mit.edu>)
Re: SQL/JSON: JSON_TABLE  (Andres Freund <andres@anarazel.de>)
List pgsql-hackers
On 3/24/22 18:54, Andrew Dunstan wrote:
> On 3/5/22 09:35, Andrew Dunstan wrote:
>> On 3/4/22 15:05, Andrew Dunstan wrote:
>>> On 3/4/22 13:13, Erikjan Rijkers wrote:
>>>> Op 04-03-2022 om 17:33 schreef Andrew Dunstan:
>>>>> This set of patches deals with items 1..7 above, but not yet the ERROR
>>>>> ON ERROR issue. It also makes some message cleanups, but there is more
>>>>> to come in that area.
>>>>>
>>>>> It is based on the latest SQL/JSON Functions patch set, which does not
>>>>> include the sql_json GUC patch.
>>>>>
>>>>> [0001-SQL-JSON-functions-without-sql_json-GUC-v56.patch]
>>>>> [0002-JSON_TABLE-v56.patch]
>>>>> [0003-JSON_TABLE-PLAN-DEFAULT-clause-v56.patch]
>>>>> [0004-JSON_TABLE-PLAN-clause-v56.patch]
>>>> Hi,
>>>>
>>>> I quickly tried the tests I already had and there are two statements
>>>> that stopped working:
>>>>
>>>> testdb=# SELECT JSON('{"a": 123, "b": [true, "foo"], "a2": "bar"}'
>>>> RETURNING jsonb);
>>>> ERROR:  syntax error at or near "RETURNING"
>>>> LINE 1: ...SON('{"a": 123, "b": [true, "foo"], "a2": "bar"}' RETURNING
>>>> ...
>>>>
>>>> testdb=# select JSON_SCALAR(123.45 returning jsonb);
>>>> ERROR:  syntax error at or near "returning"
>>>> LINE 1: select JSON_SCALAR(123.45 returning jsonb)
>>>>
>>>>   (the '^' pointer in both cases underneath 'RETURNING'
>>>>
>>>>
>>>>
>>> Yes, you're right, that was implemented as part of the GUC patch. I'll
>>> try to split that out and send new patchsets with the RETURNING clause
>>> but without the GUC (see upthread for reasons)
>>>
>>>
>> Here's a patchset with RETURNING for JSON() and JSON_SCALAR() but
>> without the GUC
>>
> Here's a new set with the latest sql/json functions patch and fixes for
> some further node handling  inadequacies.
>
>


I have been wrestling with the docs in these patches, which are somewhat
haphazardly spread across the various patches, and tying myself up in
knots. Fixing them so I don't cause later merge pain is difficult. I'm
therefore going to commit this series (staggered as previously
requested) without documentation and then commit a consolidated
documentation patch for them.


cheers


andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com




pgsql-hackers by date:

Previous
From: Justin Pryzby
Date:
Subject: Re: multithreaded zstd backup compression for client and server
Next
From: Robert Haas
Date:
Subject: Re: Assert in pageinspect with NULL pages