Re: Refactoring SysCacheGetAttr to know when attr cannot be NULL

> On 14 Mar 2023, at 08:00, Peter Eisentraut <peter(dot)eisentraut(at)enterprisedb(dot)com> wrote:

> I prefer to use "null value" for SQL null values, and NULL for the C symbol.

Thats a fair point, I agree with that.

> I'm a bit hesitant about hardcoding pg_catalog here. That happens to be true, of course, but isn't actually enforced, I think. I think that could be left off. It's not like people will be confused about which schema "pg_class.relname" is in.
>
> Also, the cached tuple isn't really for the attribute, so maybe split that up a bit, like
>
> "unexpected null value in cached tuple for catalog %s column %s"

No objections, so changed to that wording.

With these changes and a pgindent run across it per Davids comment downthread,
I've pushed this now. Thanks for review!

I'm keeping a watchful eye on the buildfarm; francolin has errored in
recoveryCheck which I'm looking into but at first glance I don't think it's
related (other animals have since passed it and it works locally, but I'll keep
digging at it to make sure).

--
Daniel Gustafsson