Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> it needs to check the tuple's cmax [...] And that means the patch
> will be a bit more invasive than this, because heap_update and
> heap_delete don't return that information at present.
I'm thinking that I could keep the test for:
GetCurrentCommandId(false) != estate->es_output_cid
as a "first pass". If that's true, I could use EvalPlanQualFetch()
to find the last version of the tuple, and generate the error if the
tuple's cmax != estate->es_output_cid. I think, although I'm not
entirely sure, that EvalPlanQualFetch needs a little work to support
this usage.
Attached is a patch based on these thoughts. Is it on the right
track? I suspect I haven't got everything covered, but thought a
reality check was in order at this point.
It does pass regression tests, including the new ones I added.
-Kevin