Hi,
On 2019-05-14 07:06:34 -0700, Andres Freund wrote:
> On May 14, 2019 4:29:01 AM PDT, Heikki Linnakangas <hlinn...@iki.fi> wrote:
> >The 'succeeded' argument seems backwards here:
> >
> >> static void
> >> heapam_tuple_complete_speculative(Relation relation, TupleTableSlot
> >*slot,
> >> uint32
> >> spekToken, bool succeeded)
> >> {
> >> bool shouldFree = true;
> >> HeapTuple tuple = ExecFetchSlotHeapTuple(slot, true, &shouldFree);
> >>
> >> /* adjust the tuple's state accordingly */
> >> if (!succeeded)
> >> heap_finish_speculative(relation, &slot->tts_tid);
> >> else
> >> heap_abort_speculative(relation, &slot->tts_tid);
> >>
> >> if (shouldFree)
> >> pfree(tuple);
> >> }
> >
> >According to the comments, if "succeeded = true", the insertion is
> >completed, and otherwise it's killed. It works, because the only caller
> >
> >is also passing the argument wrong.
>
> Thanks for finding.
>
>
> >Barring objections, I'll push the attached patch to fix that.
>
> Please hold off - your colleagues found this before, and I worked on getting
> test coverage for the code. It's scheduled for commit together today.
> Unfortunately nobody looked at the test much...
\
And pushed, as
https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=aa4b8c61d2cd57b53be03defb04d59b232a0e150
with the part that wasn't covered by tests now covered by
http://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=08e2edc0767ab6e619970f165cb34d4673105f23
Greetings,
Andres Freund
