Jeff Davis <pgsql(at)j-davis(dot)com> wrote:
> On Mon, 2011-01-31 at 15:30 -0600, Kevin Grittner wrote:
>> I'll try to set this up and see if I can get it to pass the check
>> and dcheck make targets. Can we assume that the performance
>> impact would be too small to matter when we know for sure that
> hint bits have already been set?
>
> I think that's a safe assumption. If there is some kind of
> noticeable difference in conflict rates or runtime, that probably
> indicates a bug in the new or old code.
That worked fine; passed check and dcheck targets.
Here's the code:
http://git.postgresql.org/gitweb?p=users/kgrittn/postgres.git;a=commitdiff;h=6360b0d4ca88c09cf590a75409cd29831afff58b
With confidence that it works, I looked it over some more and now
like this a lot. It is definitely more readable and should be less
fragile in the face of changes to MVCC bit-twiddling techniques. Of
course, any changes to the HTSV_Result enum will require changes to
this code, but that seems easier to spot and fix than the
alternative. Thanks for the suggestion!
Having gotten my head around it, I embellished here:
http://git.postgresql.org/gitweb?p=users/kgrittn/postgres.git;a=commitdiff;h=f9307a41c198a9aa4203eb529f9c6d1b55c5c6e1
Do those changes look reasonable? None of that is really
*necessary*, but it seemed cleaner and clearer that way once I
looked at the code with the changes you suggested.
-Kevin