Re: ON ERROR in json_query and the like

Поиск
Список
Период
Сортировка
От Amit Langote
Тема Re: ON ERROR in json_query and the like
Дата
Msg-id CA+HiwqHGVkkSde-fue+dkDxGY8_ddqBpt+pcQaAuStFzSN-MCQ@mail.gmail.com
обсуждение исходный текст
Ответ на Re: ON ERROR in json_query and the like  (Amit Langote <amitlangote09@gmail.com>)
Ответы Re: ON ERROR in json_query and the like
Список pgsql-hackers
Hi,

On Sat, Jun 22, 2024 at 5:43 PM Amit Langote <amitlangote09@gmail.com> wrote:
> On Fri, Jun 21, 2024 at 8:00 PM Markus Winand <markus.winand@winand.at> wrote:
> > So updating the three options:
> > > 1. Disallow anything but jsonb for context_item (the patch I posted yesterday)
> >
> > * Non-conforming
> > * patch available
> >
> > > 2. Continue allowing context_item to be non-json character or utf-8
> > > encoded bytea strings, but document that any parsing errors do not
> > > respect the ON ERROR clause.
> >
> > * Conforming by choosing IA050 to implement GR4: raise errors independent of the ON ERROR clause.
> > * currently committed.
> >
> > > 3. Go ahead and fix implicit casts to jsonb so that any parsing errors
> > > respect ON ERROR (no patch written yet).
> >
> > * Conforming by choosing IA050 not to implement GR4: Parsing happens later, considering the ON ERROR clause.
> > * no patch available, not trivial
> >
> > I guess I’m the only one in favour of 3 ;) My remaining arguments are that Oracle and Db2 (LUW) do it that way and
alsothat it is IMHO what users would expect. However, as 2 is also conforming (how could I miss that?), proper
documentationis a very tempting option. 
>
> So, we should go with 2 for v17, because while 3 may be very
> appealing, there's a risk that it might not get done in the time
> remaining for v17.
>
> I'll post the documentation patch on Monday.

Here's that patch, which adds this note after table 9.16.3. SQL/JSON
Query Functions:

+  <note>
+   <para>
+    The <replaceable>context_item</replaceable> expression is converted to
+    <type>jsonb</type> by an implicit cast if the expression is not already of
+    type <type>jsonb</type>. Note, however, that any parsing errors that occur
+    during that conversion are thrown unconditionally, that is, are not
+    handled according to the (specified or implicit) <literal>ON
ERROR</literal>
+    clause.
+   </para>
+  </note>

Peter, sorry about the last-minute ask, but do you have any
thoughts/advice on conformance as discussed above?

--
Thanks, Amit Langote

Вложения

В списке pgsql-hackers по дате отправления:

Предыдущее
От: Robert Haas
Дата:
Сообщение: Re: Backporting BackgroundPsql
Следующее
От: Shubham Khanna
Дата:
Сообщение: Re: Pgoutput not capturing the generated columns