Re: proposal: use errcontext for custom exception too

From: Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Robert Haas <robertmhaas(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: proposal: use errcontext for custom exception too
Date: 2011-11-25 17:49:08
Message-ID: CAFj8pRA+aryN-0kn58KN7kf1f8W1aTKntvtEdVXNrzOFmnJ5Qw@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

2011/11/25 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> On Thu, Nov 24, 2011 at 12:30 PM, Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> wrote:
>>> There are small issue in PL/pgSQL and custom exceptions. Custom
>>> exception doesn't set a CONTEXT field. I propose change this behave
>>> for WARNING or EXCEPTION level. The goal is same behave for custom
>>> exception and builtin exception and it can help to identify a RAISE
>>> statement that is responsible to exception.
>
>> That seems completely arbitrary.  I think we discussed before
>> providing an option to allow the user to control this, which seems
>> better than implementing some hardcoded rule that may or may not be
>> what a given user wants.
>
> Note also that the current behavior *is* what people want; at least,
> we have seen no field complaints about the lack of first-level CONTEXT
> for RAISE notices, and plenty of complaints from people who think
> there's still too much cruft automatically attached to RAISE output.
> If anything, what's been requested is a way to suppress even more
> context, not a policy decision to force more of it.
>

People usually don't like verbose output in interactive mode in
console. CONTEXT for RAISE NOTICE is not necessary. If you have a
small functions, then CONTEXT for RAISE EXCEPTION is not necessary
too. But if you have a functions with hundreds lines, then more
informations about origin of exception is welcome. There is workaround
- with one statement function (RAISE stmt wrapper) I have a expected
behave - but it's not clean RAISE EXCEPTION 'some message' is more
readable than PERFORM elog('some message', ..) and log is not too
readable too.

postgres=# SELECT yyy();
CONTEXT: SQL statement "SELECT xxx()"
PL/pgSQL function "yyy" line 3 at PERFORM

I can understand to motivation decrease verbosity, but there is clean
request "simply identification a source of exception (exception, not
notification)". Some RAISE stmt option should be - but for NOTICE
level NO_CONTEXT is optimal, and for EXCEPTION NO_CONTEXT is
suboptimal. It has sense just for WARNING level.

Regards

Pavel

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Jeff Janes 2011-11-25 18:52:18 Re: [PATCH] optional cleaning queries stored in pg_stat_statements
Previous Message Tom Lane 2011-11-25 16:53:21 Re: proposal: use errcontext for custom exception too