On May 25, 2010, at 4:33 PM, Marc MERLIN wrote:
> adding verify is good though, it gives a quick sanity check and it's not a
> big deal if it slows down rpm write operations which are rare for us.
False hope. There have _NEVER_ been any circumstances in RPM
where calling (*DB->verify) (or verifying Header blob
signatures/digests, another foolishness) have _EVER_
solved any "real world" problem accessing an rpmdb.
Its "stale locks", and hatred of Berkeley DB, that are
the real issues in rpmdb, not whether running (*DB->verify) (or not)
is being done. ANd "stale locks" needs a different solution,
and there's no cure for hatred of Berkeley DB.
Note that running (*DB->verify) is directly at odds with
"not happening" I/O as a goal as well.
73 de Jeff
Received on Tue May 25 22:43:10 2010