Subject: Re: jsonpath
To: Robert Haas <robertmhaas@gmail.com>
Cc: Oleg Bartunov <obartunov@gmail.com>,
 Andrew Dunstan <andrew.dunstan@2ndquadrant.com>,
 Pavel Stehule <pavel.stehule@gmail.com>,
 PostgreSQL-development <pgsql-hackers@postgresql.org>
References: <fcc6fc6a-b497-f39a-923d-aa34d0c588e8@2ndQuadrant.com>
 <CAFj8pRAhPnD=sPu-E=rwfcbgeZ8PLczYEzCsvgNYLtybSy9-=g@mail.gmail.com>
 <9f845926-0d7d-7c4d-75e3-b1e921a1677a@2ndQuadrant.com>
 <CAFj8pRBCvhgLCROfSCPGmwWZpSY+dKQN4O9xRe+1-wjBiXiHkA@mail.gmail.com>
 <bad98003-31ba-523e-d76c-3679efb2e732@2ndQuadrant.com>
 <CAFj8pRDWSCWd1JxA8srwbfL5cPgz2b5minq2UsM=cxgsFF-orw@mail.gmail.com>
 <48f13b75-0be7-c356-ff26-1db743add56e@postgrespro.ru>
 <7b4b8973-43f5-71c5-bf86-51cb5798dd88@2ndQuadrant.com>
 <f0b44fe7-d47a-eff2-592c-d9dbe4a63f75@2ndQuadrant.com>
 <5290ddd1-f23e-a53a-cae0-830dbde0a5fe@2ndQuadrant.com>
 <604ffe9c-b42e-7c42-e16b-aa44c3a12c3c@postgrespro.ru>
 <927f72f0-6832-a4ca-5676-b0ab5a2895d4@postgrespro.ru>
 <CAF4Au4wxHuEeZrMxxKXUYcs1kDws-f0oW0-sEyun388-U-NOxw@mail.gmail.com>
 <9a1ac993-7608-797c-2095-c735605da82f@postgrespro.ru>
 <CA+TgmoYXtoPPaV-OEUfqujCcfKp+ZT=KOuPHkhRoQL_iYzQ8yQ@mail.gmail.com>
From: Nikita Glukhov <n.gluhov@postgrespro.ru>
Organization: Postgres Professional
Message-ID: <94ace92c-714c-903c-fab2-31fb838204ad@postgrespro.ru>
Date: Fri, 2 Mar 2018 00:40:51 +0300
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.6.0
MIME-Version: 1.0
In-Reply-To: 
 <CA+TgmoYXtoPPaV-OEUfqujCcfKp+ZT=KOuPHkhRoQL_iYzQ8yQ@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Language: en-US
Precedence: bulk

On 28.02.2018 06:55, Robert Haas wrote:

> On Mon, Feb 26, 2018 at 10:34 AM, Nikita Glukhov
> <n.gluhov@postgrespro.ru> wrote:
>> Attached 10th version of the jsonpath patches.
>>
>> 1. Fixed error handling in arithmetic operators.
>>
>>     Now run-time errors in arithmetic operators are catched (added
>>     PG_TRY/PG_CATCH around operator's functions calls) and converted into
>>     Unknown values in predicates as it is required by the standard:
> I think we really need to rename PG_TRY and PG_CATCH or rethink this
> whole interface so that people stop thinking they can use it to
> prevent errors from being thrown.

I understand that it is unsafe to call arbitrary function inside PG_TRY without
rethrowing of caught errors in PG_CATCH, but in jsonpath only the following
numeric and datetime functions with known behavior are called inside PG_TRY
and only errors of category ERRCODE_DATA_EXCEPTION are caught:

numeric_add()
numeric_mul()
numeric_div()
numeric_mod()
numeric_float8()
float8in()
float8_numeric()
to_datetime()

SQL/JSON standard requires us to handle errors and then perform the specified
ON ERROR behavior.  In the next SQL/JSON patch I had to use subtransactions for
catching errors in JSON_VALUE, JSON_QUERY and JSON_EXISTS where an arbitrary
user-defined typecast function can be called.

-- 
Nikita Glukhov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company