Re: CLOBBER_CACHE_ALWAYS regression instability - Mailing list pgsql-hackers

From Andres Freund
Subject Re: CLOBBER_CACHE_ALWAYS regression instability
Date
Msg-id 20200405204404.jw7yq4p5kmdna2a4@alap3.anarazel.de
Whole thread Raw
In response to Re: CLOBBER_CACHE_ALWAYS regression instability  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On 2020-04-05 15:38:29 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > We could print the error using :LAST_ERROR_MESSAGE after removing a
> > potential trailing "at character ..." if we're worried about the loss of
> > specificity.
> 
> Oh, actually it seems that :LAST_ERROR_MESSAGE is already just the
> primary message, without any "at character N" addon, so this would be
> a very easy way to ameliorate that complaint.  ("at character N" is
> added by libpq's pqBuildErrorMessage3 in TERSE mode, but psql does
> not use that when filling LAST_ERROR_MESSAGE.)

Heh. I though it worked differently because I just had typed some
gibberish and got an error in :LAST_ERROR_MESSAGE that ended in "at or
near ...". But that's scan.l adding that explicitly...



pgsql-hackers by date:

Previous
From: Justin Pryzby
Date:
Subject: Re: VACUUM memory management
Next
From: Alvaro Herrera
Date:
Subject: Re: Add A Glossary