On Mon, May 2, 2011 at 3:37 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Simon Riggs <simon@2ndQuadrant.com> writes:
>> I can't see the objection to replacing something inadvertently removed
>> in 9.0, especially since it is a 1 line patch and is accompanied by
>> copious technical evidence.
>
> I am not sure which part of "this isn't a substitute for what happened
> before 9.0" you fail to understand.
>
> As for "copious technical evidence", I saw no evidence provided
> whatsoever that this patch really did anything much to fix the
> reported problem.
Just so we're looking at the same data, graph attached.
> Yeah, it would help during the initial scan
> of the old rel, but not during the sort or reindex steps.
As Greg points out, the sort is not really of concern (for now).
> (And as for the thoroughness of the technical analysis, the patch
> doesn't even catch the second CHECK_FOR_INTERRUPTS in copy_heap_data;
> which would at least provide some relief for the sort part of the
> problem, though only in the last pass of sorting.)
I'm sure Gabriele can add those things as well - that also looks like
another 1 line change.
I'm just observing that the patch as-is appears effective and I feel
it is important.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
