Re: Do we want a hashset type? - Mailing list pgsql-hackers

From Joel Jacobson
Subject Re: Do we want a hashset type?
Date
Msg-id 0a87fe7e-e0d9-4d7b-8c5a-6d93d62fd542@app.fastmail.com
Whole thread Raw
In response to Re: Do we want a hashset type?  (Tomas Vondra <tomas.vondra@enterprisedb.com>)
Responses Re: Do we want a hashset type?
List pgsql-hackers
On Tue, Jun 20, 2023, at 14:10, Tomas Vondra wrote:
> This is also what the SQL standard does for multisets - there's SQL:20nn
> draft at http://www.wiscorp.com/SQLStandards.html, and the <member
> predicate> section (p. 475) explains how this should work with NULL.

I've looked again at the paper you mentioned and found something intriguing
in section 2.6 (b). I'm a bit puzzled about this: why would we want to return
null when we're certain it's not null but just doesn't have any elements?

In the same vein, it says, "If it has more than one element, an exception is
raised." Makes sense to me, but what about when there are no elements at all?
Why not raise an exception in that case too?

The ELEMENT function is designed to do one simple thing: return the element of
a multiset if the multiset has only 1 element. This seems very similar to how
our INTO STRICT operates, right?

The SQL:20nn seems to still be in draft form, and I can't help but wonder if we
should propose a bit of an improvement here:

"If it doesn't have exactly one element, an exception is raised."

Meaning, it would raise an exception both if there are more elements,
or zero elements (no elements).

I think this would make the semantics more intuitive and less surprising.

/Joel



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [PATCH] Extend ALTER OPERATOR to support adding commutator, negator, hashes, and merges
Next
From: Pavel Stehule
Date:
Subject: Re: Schema variables - new implementation for Postgres 15