On Wed, 2013-09-04 at 11:32 -0400, Tom Lane wrote:
> Jeff Davis <pgsql@j-davis.com> writes:
> > I think code complexity matters quite a lot. If we can eliminate some
> > complex code in a complex area, and all we give up is a feature with
> > essentially no use case, that sounds like we're moving in the right
> > direction to me.
>
> Isn't this whole discussion academic in view of Andres' point?
Maybe "complex code" was an overstatement. We'd be able to eliminate the
XLOG_FPW_CHANGE, UpdateFullPageWrites(), and one of the members of
XLogCtlInsert; and make xlog.c slightly shorter in the process.
The first time I looked at doing the patch to honor full_page_writes=off
when checksums are on, the fact that fullPageWrites was changeable was a
distraction. Since I saw little or no value in what the code offered, my
instinct was to see if we could get rid of it.
It looks like Simon went to significant effort to maintain the
full_page_writes as a PGC_SUGHUP:
http://git.postgresql.org/gitweb/?p=postgresql.git;a=commitdiff;h=8366c780
Maybe he has the best perspective on the value versus complexity?
Regards,Jeff Davis