Re: pgsql: Avoid pin scan for replay of XLOG_BTREE_VACUUM

On 2016-01-09 17:58:01 +0000, Simon Riggs wrote:
> On 9 January 2016 at 12:23, Andres Freund <andres(at)anarazel(dot)de> wrote:
>
> > Hi,
> >
> > On 2016-01-09 10:13:27 +0000, Simon Riggs wrote:
> > > src/backend/access/rmgrdesc/nbtdesc.c | 2 +-
> >
> > I've not reviewed the patch, but a very quick glance during a rebase
> > with conflicts showed:
> >
> > @@ -48,7 +48,7 @@ btree_desc(StringInfo buf, XLogReaderState *record)
> > {
> > xl_btree_vacuum *xlrec = (xl_btree_vacuum *) rec;
> >
> > - appendStringInfo(buf, "lastBlockVacuumed %u",
> > + appendStringInfo(buf, "lastBlockVacuumed %d",
> > xlrec->lastBlockVacuumed);
> > break;
> > }
> >
> > which doesn't look right?
> >
>
> It's right. New value of -1 allowed in that field, so change required to
> allow it to display properly for debug.

Uh. xl_btree_vacuum->lastBlockVacuumed is of type BlockNumber, which in
turn is of type uint32. So no, this isn't correct as is.