Re: datatype of constant is not propagated into aggregate query

From: Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: datatype of constant is not propagated into aggregate query
Date: 2012-03-11 15:50:05
Message-ID: CAFj8pRCXvyvYFDqzDbBYm+04SCgc3nZfr_P5nh+GOnm8wAM3Fw@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

2012/3/11 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>:
> Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> writes:
>> postgres=# insert into t1 select distinct '2001-01-01', 1 from
>> generate_series(1,3);
>> ERROR:  column "d" is of type date but expression is of type text
>> LINE 1: insert into t1 select distinct '2001-01-01', 1 from generate...
>>                                        ^
>> HINT:  You will need to rewrite or cast the expression.
>
> The DISTINCT forces the sub-SELECT to decide on a type for the constant
> (so that it can pick a semantics for DISTINCT-ing).  And it chooses text
> by default.  This is maybe not terribly convenient, but I'm not sure it
> would be a good idea at all for a surrounding INSERT to change the
> semantics of a SELECT.  Even discounting the implementation
> difficulties, I don't think that'd satisfy the POLA.
>
> (We do have a hack for adjusting the type if what the sub-select returns
> is still of type UNKNOWN; that's safe because it implies that the
> sub-select does not care about the type of the result column.)

in this case the constant is forwarded to result without any change,
so it can be UNKNOWN.

Pavel

>
>                        regards, tom lane

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2012-03-11 16:00:47 Re: NULL's support in SP-GiST
Previous Message Pavel Stehule 2012-03-11 15:45:35 Re: datatype of constant is not propagated into aggregate query