"Euler Taveira" <euler@eulerto.com> writes:
> On Wed, Aug 2, 2023, at 5:53 PM, Euler Taveira wrote:
>> On Wed, Aug 2, 2023, at 5:06 PM, Tom Lane wrote:
>>> Yeah, ignoring dependencies on ACLs in this logic seems like the best
>>> way forward. Do you want to write a patch?
>> I like your suggestion. Let me give it a try.
Hmm, changing it like that seems likely to have a lot of unexpected
side-effects. What I had in mind was to change the
if (te->nDeps != 1 ||
TocIDRequired(AH, te->dependencies[0]) == 0)
return 0;
bit so that it would loop through the TE's dependencies to see
if any one of them is a required non-ACL TE. We have to relax
the "te->nDeps != 1" restriction, but we still want to check
that there is a parent object that is being restored.
Some work on the associated comment block seems appropriate too.
And maybe add a test case?
regards, tom lane