Date: Thu, 16 May 2019 10:43:08 +0200 (CEST)
From: Fabien COELHO <coelho@cri.ensmp.fr>
To: Daniel Verite <daniel@manitou-mail.org>
cc: "Iwata, Aya" <iwata.aya@jp.fujitsu.com>,
    PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>
Subject: RE: psql - add SHOW_ALL_RESULTS option
In-Reply-To: <710caf32-a5f9-4511-bd9d-736c1a91f4c3@manitou-mail.org>
Message-ID: <alpine.DEB.2.21.1905161037270.19737@lancre>
References: <710caf32-a5f9-4511-bd9d-736c1a91f4c3@manitou-mail.org>
User-Agent: Alpine 2.21 (DEB 202 2017-01-01)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII; format=flowed
Precedence: bulk


Bonjour Daniel,

>>>> IMHO this new setting should be on by default: few people know about \; so
>>>> it would not change anything for most, and I do not see why those who use
>>>> it would not be interested by the results of all the queries they asked for.
>>> I agree with your opinion.
>>
>> Ok. I did not yet change the default in the attached version, though.
>
> I'd go further and suggest that there shouldn't be a variable
> controlling this. All results that come in should be processed, period.
> It's not just about \; If the ability of CALL to produce multiple
> resultsets gets implemented (it was posted as a POC during v11
> development), this will be needed too.

I do agree, but I'm afraid that if there is no opt-out it could be seen as 
a regression by some.

>> This attached version does:
>>  - ensure that warnings appear just before its
>>  - add the entry in psql's help
>>  - redefine the function boundary so that timing is cleaner
>>  - include somehow improved tests
>
> \errverbose seems to no longer work with the patch:
>
> test=> select 1/0;
> psql: ERROR:  division by zero
>
> test=> \errverbose
> There is no previous error.
>
> as opposed to this output with PG11:
>
> test=> \errverbose
> ERROR:	22012: division by zero
> LOCATION:  int4div, int.c:820

Thanks for the catch. I'll investigate.

> \errverbose has probably no regression tests because its output includes 
> these ever-changing line numbers; hence `make check` cannot be used to 
> find this regression.

What is not tested does not work:-( The TAP infrastructure for psql 
included in some patch (https://commitfest.postgresql.org/23/2100/ I 
guess) would help testing such slightly varying features which cannot be 
tested with a hardcoded reference text.

-- 
Fabien.