Hannu Krosing wrote:
> On Wed, 2009-04-15 at 18:04 -0400, Tom Lane wrote:
>
>> Simon Riggs <simon@2ndQuadrant.com> writes:
>>
>>> I've always been scared to ask this question, in case the answer is No,
>>> but: Do we have a set of regression tests for the optimizer anywhere?
>>>
>> Nothing beyond what is in the standard tests. While that's okay at
>> catching wrong answers --- and we have memorialized a number of such
>> issues in the tests --- the framework is not good for catching things
>> that run slower than they ought.
>>
>
> Can't we make first cut at it by just running with timings on and then
> compare ratios of running times - maybe with 2-3X tolerance - to catch
> most obvious regressions ?
>
>
The current regression tests are a series of yes/no answers to this
question: does the actual output match the expected output. Nothing like
as fuzzy as what you are suggesting is supported at all. From time to
time suggestions are made for a performance farm as a kind of analog to
the buildfarm, which would look at quantitative timing tests rather than
just success/failure tests. It on my (very long) list of things to do,
but it not something we can just tack on to the current regression suite
simply.
cheers
andrew