Re: Remove page-read callback from XLogReaderState.

At Thu, 30 Sep 2021 09:40:06 +0900 (JST), Kyotaro Horiguchi <horikyota(dot)ntt(at)gmail(dot)com> wrote in
> At Mon, 27 Sep 2021 17:31:03 +1300, Thomas Munro <thomas(dot)munro(at)gmail(dot)com> wrote in
> > On Thu, Jul 15, 2021 at 4:48 PM Kyotaro Horiguchi
> > <horikyota(dot)ntt(at)gmail(dot)com> wrote:
> > > Gah... Thank you for noticing me. I thought that I have sent the
> > > rebased version. This is the rebased version on the current master.
> >
> > Hi Kyotaro,
> >
> > Did you see this?
> >
> > https://www.postgresql.org/message-id/20210429022553.4h5qii5jb5eclu4i%40alap3.anarazel.de
>
> Thank you for pinging me. I haven't noticed of that.
> I'll check on that line.

It looks like the XLogFindNextRecord was not finished. It should have
been turned into a state machine.

In this version (v18),

This contains only page-reader refactoring stuff.

- Rebased to the current master, including additional change for
XLOG_OVERWRITE_CONTRECORD stuff. (This needed the new function
XLogTerminateRead.)

- Finished XLogFindNextRecord including the fixup from Thomas' v17.

- Added a test for XLogFindNextRecord, on the behavior that
page-skipping on seeking for the first record.

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center