Re: relhassubclass and partitioned indexes

Thanks for commenting.

On 2018/10/19 15:17, Michael Paquier wrote:
> On Fri, Oct 19, 2018 at 01:45:03AM -0400, Tom Lane wrote:
>> Amit Langote <Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp> writes:
>>> Should relhassubclass be set/reset for partitioned indexes?
>>
>> Seems like a reasonable idea to me, at least the "set" end of it.
>> We don't ever clear relhassubclass for tables, so maybe that's
>> not necessary for indexes either.

We *do* reset opportunistically during analyze of inheritance parent; the
following code in acquire_inherited_sample_rows() can reset it:

/*
* Check that there's at least one descendant, else fail. This could
* happen despite analyze_rel's relhassubclass check, if table once had a
* child but no longer does. In that case, we can clear the
* relhassubclass field so as not to make the same mistake again later.
* (This is safe because we hold ShareUpdateExclusiveLock.)
*/
if (list_length(tableOIDs) < 2)
{
/* CCI because we already updated the pg_class row in this command */
CommandCounterIncrement();
SetRelationHasSubclass(RelationGetRelid(onerel), false);
ereport(elevel,
(errmsg("skipping analyze of \"%s.%s\" inheritance tree ---
this inheritance tree contains no child tables",
get_namespace_name(RelationGetNamespace(onerel)),
RelationGetRelationName(onerel))));
return 0;
}

Similarly, perhaps we can make REINDEX on a partitioned index reset the
flag if it doesn't find any children. We won't be able to do that today
though, because:

static void
ReindexPartitionedIndex(Relation parentIdx)
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("REINDEX is not yet implemented for partitioned
indexes")));
}

> No objections to the proposal. Allowing find_inheritance_children to
> find index trees for partitioned indexes could be actually useful for
> extensions like pg_partman.

Thanks. Attached a patch to set relhassubclass when an index partition is
added to a partitioned index.

Regards,
Amit