Re: Cache lookup errors with functions manipulation object addresses

> On 17 Oct 2019, at 03:37, Michael Paquier <michael(at)paquier(dot)xyz> wrote:

> Attached is an updated
> patch set with the heap_close() calls removed as per the recent report
> from Dmitry.

Taking a look at this to see if we can achieve closure on this long-running
patchset. The goal of the patch is IMO a clear win for usability.

The patchset applies with a bit of fuzzing and offsetting, it has tests (which
pass) as well as the relevant documentation changes. I agree with the previous
reviewers that the new shape of the test is better, so definitely +1 on that
change. There are a lot of mechanic changes in this set, but AFAICT they cover
all the bases.

Since the patch has been through a lot of review already there isn't a lot to
say, only a few very minor things that stood out:

* This exports the useful functionality of regoperatorout for use
* in other backend modules. The result is a palloc'd string.
format_operator_extended has this comment, but the code can now also return
NULL and not just a palloc'd string.

+ if (!missing_ok)
+ {
+ elog(ERROR, "could not find tuple for cast %u",
+ object->objectId);
+ }
The other (!missing_ok) blocks in this patch are not encapsulating the elog()
with braces.

I'm switching this to ready for committer,

cheers ./daniel