On Fri, Jan 27, 2012 at 4:07 PM, Dan Scales <scales@vmware.com> wrote:
> The advantage of putting the checksum calculation in smgrwrite() (or mdwrite()) is that it catches a bunch of page
writesthat don't go through the buffer pool (see calls to smgrwrite() in nbtree.c, nbtsort.c, spginsert.c)
Maybe we should have some sort of wrapper function, then. I really
dislike the idea that the smgr layer knows anything about the page
format, and if md has to know that's even worse.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company