On 2021/07/13 5:57, Matthias van de Meent wrote:
>
>
> On Mon, 12 Jul 2021 at 13:14, <ikedamsh@oss.nttdata.com
> <mailto:ikedamsh@oss.nttdata.com>> wrote:
>>
>> Hi,
>>
>> While I’m reading source codes related to vacuum, I found comments which
>> don’t seem to fit the reality. I think the commit[1] just forgot to fix them.
>> What do you think?
>
> Hmm, yes, those are indeed some leftovers.
>
> Some comments on the suggested changes:
>
>
> - * caused by HeapTupleSatisfiesVacuum. We just add entries to the arrays in
> + * caused by heap_prune_satisfies_vacuum. We just add entries to the arrays in
>
> I think that HeapTupleSatisfiesVacuumHorizon might be an alternative correct
> replacement here.
>
>
> - elog(ERROR, "unexpected HeapTupleSatisfiesVacuum result");
> + elog(ERROR, "unexpected heap_prune_satisfies_vacuum result");
>
> The type of the value is HTSV_Result; where HTSV stands for
> HeapTupleSatisfiesVacuum, so if we were to replace this, I'd go for
> "unexpected result from heap_prune_satisfies_vacuum" as a message instead.
Thanks for your comments. I agree your suggestions.
I also updated prstate->vistest to heap_prune_satisfies_vacuum of v1 patch
because heap_prune_satisfies_vacuum() tests with not only prstate->vistest but
also prstate->old_snap_xmin. I think it's more accurate representation.
Regards,
--
Masahiro Ikeda
NTT DATA CORPORATION