On Mon, Nov 22, 2021 at 11:21:52AM +0100, Peter Eisentraut wrote:
> On 19.11.21 08:31, Michael Paquier wrote:
>> Regarding 0001, I find the existing code a bit more self-documenting
>> if we keep those checks flagInhAttrs() and guessConstraintInheritance().
>> So I would rather leave these.
>
> In that case, the existing check in guessConstraintInheritance() seems
> wrong, because it doesn't check for RELKIND_MATVIEW. Should we fix that?
> It's dead code either way, but if the code isn't exercises, then these kinds
> of inconsistency come about.
Yeah, this one could be added. Perhaps that comes down to one's taste
at the end, but I would add it.
> Maybe
>
> else
> {
> Assert(RELKIND_HAS_STORAGE(rel->rd_rel->relkind);
> RelationCreateStorage(rel->rd_node, relpersistence);
> }
>
> create_storage is set earlier based on RELKIND_HAS_STORAGE(), so this would
> be consistent.
Sounds fine by me. Perhaps you should apply the same style in
RelationGetNumberOfBlocksInFork(), then?
--
Michael