Re: to be a multirange or not be, that's the question - Mailing list pgsql-hackers

From David G. Johnston
Subject Re: to be a multirange or not be, that's the question
Date
Msg-id CAKFQuwZLmmpaB_08+vE48YLSbk6e+tGCrLUh+rE2D2Q9X_kL_w@mail.gmail.com
Whole thread Raw
In response to to be a multirange or not be, that's the question  (Jaime Casanova <jcasanov@systemguards.com.ec>)
List pgsql-hackers
On Saturday, November 6, 2021, Jaime Casanova <jcasanov@systemguards.com.ec> wrote:
Ok, subject was a bit philosophical but this message I just found is
quite confusing.

"""
regression=# select cast(null as anyrange) &> cast(null as anymultirange);
ERROR:  argument declared anymultirange is not a multirange type but type anymultirange
"""


I get that it is hard to parse but it is unambiguously correct.  It does seem to presume that one understands how polymorphic pseudo-types work (the supplied query was written without that knowledge applied).  I can envision some improvements here though it would depend a lot on the actual implementation.  I’m also curious as to why we get as far the operator invocation when casting literals to polymorphic pseudo-types is supposedly an invalid thing to do.

David J.

pgsql-hackers by date:

Previous
From: Jaime Casanova
Date:
Subject: to be a multirange or not be, that's the question
Next
From: Pavel Stehule
Date:
Subject: Re: [PROPOSAL] new diagnostic items for the dynamic sql