|From:||"Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov>|
|To:||<drkp(at)csail(dot)mit(dot)edu>,<heikki(dot)linnakangas(at)enterprisedb(dot)com>, "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov>|
|Subject:||Re: Repeated PredicateLockRelation calls during seqscan|
|Views:||Raw Message | Whole Thread | Download mbox | Resend email|
"Kevin Grittner" wrote:
> "Kevin Grittner" wrote:
>> Heikki Linnakangas wrote:
>>> BTW, isn't bitgetpage() in nodeBitmapHeapscan.c missing
>>> PredicateLockTuple() and CheckForSerializableConflictOut() calls
>>> in the codepath for a lossy bitmap? In the non-lossy case,
>>> heap_hot_search_buffer() takes care of it, but not in the lossy
>> I think the attached addresses that.
> Don't commit that patch, it's not holding up in testing here.
> I'll look at it some more.
Version 2 is attached. It initializes some data which was
uninitialized in a HeapTableData structure which already existed in
the code. I've been burned by this before -- making a seemingly
innocuous change to code which then fails because the comments at
lines 512 to 514 in htup.h are not actually true:
I asked about this the first time it bit me in this thread:
which concluded here:
Having been bitten by it a *second* time now, I'm inclined to go
through and make the code match the comments wherever these
structures are used. It's a bit late in the cycle to do that for
9.1, but I'll get something on the table for 9.2 if nobody wants to
argue against that course.
|Next Message||Peter Eisentraut||2011-06-26 22:24:40||Re: pg_upgrade defaulting to port 25432|
|Previous Message||Kevin Grittner||2011-06-26 20:02:12||Re: Repeated PredicateLockRelation calls during seqscan|