Re: Allowing join removals for more join types

On 23 June 2014 12:06, David Rowley <dgrowley(at)gmail(dot)com> wrote:

>> It's not clear to me where you get the term "sortclause" from. This is
>> either the groupclause or distinctclause, but in the test cases you
>> provide this shows this has nothing at all to do with sorting since
>> there is neither an order by or a sorted aggregate anywhere near those
>> queries. Can we think of a better name that won't confuse us in the
>> future?
>>
>
> I probably got the word "sort" from the function targetIsInSortList, which
> expects a list of SortGroupClause. I've renamed the function to
> sortlist_is_unique_on_restrictinfo() and renamed the sortclause parameter to
> sortlist. Hopefully will reduce confusion about it being an ORDER BY clause
> a bit more. I think sortgroupclauselist might be just a bit too long. What
> do you think?

OK, perhaps I should be clearer. The word "sort" here seems completely
misplaced and we should be using a more accurately descriptive term.
It's slightly more than editing to rename things like that, so I'd
prefer you cam up with a better name.

Did you comment on the transitive closure question? Should we add a
test for that, whether or not it works yet?

Other than that it looks pretty good to commit, so I'll wait a week
for other objections then commit.

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