Re: simplifying emode_for_corrupt_record

Robert Haas <robertmhaas@gmail.com>

From: Robert Haas <robertmhaas@gmail.com>

To: Tom Lane <tgl@sss.pgh.pa.us>

Cc: Bruce Momjian <bruce@momjian.us>, Magnus Hagander <magnus@hagander.net>, Heikki Linnakangas <heikki.linnakangas@enterprisedb.com>, Simon Riggs <simon@2ndquadrant.com>, Fujii Masao <masao.fujii@gmail.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>

Date: 2010-06-28T19:42:52Z

Lists: pgsql-hackers

On Fri, Jun 25, 2010 at 5:07 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> I spend a little bit of time analyzing this today and it appears to me
> that all of the calls to emode_for_corrupt_record() arrive via
> ReadRecord(), which itself takes an emode argument that is always
> passed by the caller as either LOG or PANIC.  Therefore, the effect of
> the first "if" test in emode_for_corrupt_record() is to reduce the
> logging level of messages coming from SR or the archive from LOG to
> WARNING.  (WARNING would be higher in an interactive session, but not
> here, per Tom's point.)  This seems clearly a bad idea, so I propose
> to rip it out, which simplifies this function considerably.  Proposed
> patch attached.

Since this appears to be non-controversial, I'm going to go ahead and commit it.

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