Greetings,
* Kyotaro HORIGUCHI (horiguchi.kyotaro@lab.ntt.co.jp) wrote:
> At Wed, 28 Nov 2018 10:01:36 +0900, Michael Paquier <michael@paquier.xyz> wrote in
<20181128010136.GU1716@paquier.xyz>
> > On Tue, Nov 27, 2018 at 07:51:03PM -0500, Stephen Frost wrote:
> > > That isn't at all what I got from that.
> > >
> > > A rewrite of this really should avoid talking about removing the hole as
> > > if it's 'compression' because, first of all, it isn't, and second, now
> > > that we have *actual* compression happening, it's terribly confusing to
> > > talk about removing the hole using that terminology.
> >
> > You may want to be careful about other comments in xloginsert.c or such
> > then. Removing a page hole is mentioned in a couple of places as a form
> > of compression if I recall correctly.
>
> org> When wal_compression is enabled, a full page image which "hole" was
> org> removed is additionally compressed using PGLZ compression algorithm.
>
> Even though it has an obvious mistake, I could read this as
> full_page_image is always compressed after removing a hole in it
> if any. (I'm not sure it makes correct sense, though.) I feel
> hopelessly that the sentence structure model in my mind is
> different from that of natives. I need to study harder, but..
Thanks to everyone for sharing their thoughts here, the goal is simply
to try and have the comments as clear as we can for everyone.
Please find attached a larger rewording of the comment in xlogrecord.h,
and a minor change to xloginsert.c to clarify that we're removing a hole
and not doing compression at that point.
Other thoughts on this..?
Thanks!
Stephen
