Re: hot update doesn't work?

On Wed, May 12, 2010 at 11:47 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> "Kevin Grittner" <Kevin.Grittner@wicourts.gov> writes:
>> You're updating the row 100000 times within a single transaction.  I
>> don't *think* HOT will reclaim a version of a row until the
>> transaction which completed it is done and no other transactions can
>> see that version any longer.  It does raise the question, though --
>> couldn't a HOT update of a tuple *which was written by the same
>> transaction* do an "update in place"?
>
> Well ... in the first place there is not, ever, any such thing as
> "update in place".  The correct question to ask is whether we could
> vacuum away the older elements of the HOT chain on the grounds that they
> are no longer of interest.  What we would see is tuples with xmin equal
> to xmax and cmin different from cmax.  The problem then is to determine
> whether there are any live snapshots with curcid between cmin and cmax.
> There is 0 hope of doing that from outside the originating backend.
> Now if heap_page_prune() is being run by the same backend that generated
> the in-doubt tuples, which I will agree is likely in a case like this,
> in principle we could do it.  Not sure if it's really worth the trouble
> and nonorthogonal behavior.

update of same row in a single transaction is not going to come up
that much and there are a number of simple work arounds to get better
performance.

isn't it possible to skip the snapshot check for temp tables though?

merlin