Re: Suspicious behaviour on applying XLOG_HEAP2_VISIBLE.

On 2016-04-24 22:10:19 +0100, Simon Riggs wrote:
> On 18 April 2016 at 13:15, Simon Riggs <simon@2ndquadrant.com> wrote:
> > (non-reply just because of travel)
> >
> > OK, I'll write up a patch today to fix, with a view to backpatching.
> >
> 
> Patch from Tuesday.  On various planes.

I posted a version of this at
http://archives.postgresql.org/message-id/20160424025117.2cmf6ku4ldfcoo44%40alap3.anarazel.de
(because it was blocking me from fixing a newer bug)

I think my approach of a separate record is going to be easier to
backpatch - the new commit record format you took advantage of is
new. If we go your way, this also needs pg_xlogdump/xact_desc() &
logical decoding integration.

Greetings,

Andres Freund