Re: Do we need so many hint bits?

On 15 November 2012 22:08, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Simon Riggs <simon(at)2ndQuadrant(dot)com> writes:
>> On 15 November 2012 19:42, Jeff Davis <pgsql(at)j-davis(dot)com> wrote:
>>> many of the hint bits aren't terribly important
>
>> The truth is that nobody knows because there is no way of knowing.
>
> We had a discussion awhile back in which the idea of *no* hint bits
> was advocated, and someone (I think Robert) did some preliminary
> tests that pretty much shot it down. However, I don't recall
> anyone suggesting before that the four existing bits might not all
> be equally worthwhile. It's worth looking into.

Itsn't that what I said? In case of doubt, Yes, its a great idea and
worth looking into.

The question is *how* we look into it.

> The hard part is
> probably agreeing on the test case or cases to measure behavior for.

I think thats impossible. There are just too many possible cases.
Measuring top-level performance without measuring low level stats just
means we can't tell the difference between a test that didn't exercise
the code and a test where there was no difference.

We need detailed stats that allow many people to make their own tests
and to report on what they find.

--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services