Dear Heikki,
> Hmm. ResOwnerReleaseTupleDesc() does exist, those functions are needed
> for the callbacks. I think you meant the wrappers around
> ResourceOwnerRemember and ResourceOwnerForget, like
> ResourceOwnerRememberCatCacheRef(). I admit those are not fully
> consistent: I didn't bother with the wrapper functions when there is
> only one caller.
Yeah, I meant it. And I prefer your policy.
> Hmm. ResOwnerReleaseTupleDesc() does exist, those functions are needed
> for the callbacks. I think you meant the wrappers around
> ResourceOwnerRemember and ResourceOwnerForget, like
> ResourceOwnerRememberCatCacheRef(). I admit those are not fully
> consistent: I didn't bother with the wrapper functions when there is
> only one caller.
Good job. I confirmed your fixes, and I confirmed it looks fine.
I will check another ResourceOwnerEnlarge() if I have a time.
> I've been working on performance testing too.
I'm looking forward to seeing it.
Best regards,
Hayato Kuroda
FUJITSU LIMITED