Re: psql's ON_ERROR_STOP is misdocumented

On Mon, Jun 13, 2011 at 11:01 AM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Thu, May 5, 2011 at 11:20 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> The documentation for ON_ERROR_STOP states, or at least implies by
>> omission, that it only affects the behavior in non-interactive scripts:
>>
>>        By default, if non-interactive scripts encounter an error, such
>>        as a malformed SQL command or internal meta-command, processing
>>        continues. This has been the traditional behavior of psql but it
>>        is sometimes not desirable. If this variable is set, script
>>        processing will immediately terminate. If the script was called
>>        from another script it will terminate in the same fashion. If
>>        the outermost script was not called from an interactive psql
>>        session but rather using the -f option, psql will return error
>>        code 3, to distinguish this case from fatal error conditions
>>        (error code 1).
>>
>> However, it is easily proven that it *does* affect interactive commands;
>> just try two commands on one line:
>>
>> regression=# select 1/0; select 2;
>> ERROR:  division by zero
>>  ?column?
>> ----------
>>        2
>> (1 row)
>>
>> regression=# \set ON_ERROR_STOP 1
>> regression=# select 1/0; select 2;
>> ERROR:  division by zero
>> regression=#
>>
>> Can we get the docs changed to reflect reality?
>
> Here's an attempt at some suitable word-smithing.

Hearing no objections, committed.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company