Some illogical query results

[dmg2206]

I've gotten the following:

db=# select count(*) from table1 where field1 in (select field2 from table2);
 count
-------
   175
(1 row)

db=# select count(*) from table1 where field1 not in (select field2 from table2);
 count
-------
     0
(1 row)

db=# select count(*) from table1;
 count
-------
   955
(1 row)

OK, isn't this mathematically impossible? Any idea what's going on here?

 

[rycamor]

Yes, most likely you have some NULLs involved here. Aggregate functions such as COUNT(*) cannot perform certain kinds of logic with a NULL, because NULL doesn't mean "no value"; in relational query terms it means "we don't know if there is a value", thus it can't logically tell whether field1 does or does not match field2. This is why many serious DB practitioners will caution you to avoid creating columns that allow NULLs unless you have a specific reason to do so. -------------------------------------------

Big Brother: "War is Peace" -- Big Business: "Trust is Suspicion"
(

http://www.cl.cam.ac.uk/~rja14/tcpa-faq.html)

 

[dmg2206]

Now I've got a new problem. I've set table2 to mirror table1 in structure, and now I've dumped new data into table2. I'm trying to insert all the data from table2 into table1, but I keep getting this error:

ERROR:  Cannot insert a duplicate key into unique index indx_unique_table1_field

indx_unique_table1_field is a unique index I set up on the column "field" in table1. However, when I search for these duplicate values, I get the following:

select field from table2 where field in (select field from table1);
 field
-------
(0 rows)

There are no null values in either of the columns in question.

 

[rycamor]

How about showing us the structure of both tables? (

pg_dump -s

) -------------------------------------------

Big Brother: "War is Peace" -- Big Business: "Trust is Suspicion"
(

http://www.cl.cam.ac.uk/~rja14/tcpa-faq.html)

 

[dmg2206]

Here's the two tables:

CREATE TABLE "tbclients" (
	"id" integer DEFAULT nextval('"seq_tbclients_id"'::text) NOT NULL,
	"ssn" character varying(11),
	"firstname" character varying(20),
	"lastname" character varying(30),
	"street" character varying(50),
	"city" character varying(25),
	"state" character varying(2) DEFAULT 'IN',
	"zipcode" integer,
	"phone" character varying(14),
	"dob" date,
	"sex" character varying(6),
	"township" character varying(20),
	"veteran" boolean,
	"ethnicity" character varying(20),
	"maritalstatus" character varying(20),
	"residentdate" date,
	"church" character varying(50),
	"insurance" character varying(50),
	"resident" character varying(15),
	"homeless" boolean,
	Constraint "tbclients_pkey" Primary Key ("id")
);


CREATE TABLE "ccsnclients" (
	"id" integer DEFAULT nextval('"seq_ccsnclients_id"'::text) NOT NULL,
	"ssn" character varying(11),
	"firstname" character varying(20),
	"lastname" character varying(30),
	"street" character varying(50),
	"city" character varying(25),
	"state" character varying(2) DEFAULT 'IN',
	"zipcode" integer,
	"phone" character varying(14),
	"dob" date,
	"sex" character varying(6),
	"township" character varying(20),
	"veteran" boolean,
	"ethnicity" character varying(20),
	"maritalstatus" character varying(20),
	"residentdate" date,
	"church" character varying(50),
	"insurance" character varying(50),
	"resident" character varying(15),
	"homeless" boolean,
	Constraint "ccsnclients_pkey" Primary Key ("id")
);

I'm trying to insert data from ccsnclients into tbclients. I use the SQL command

insert into tbclients (select * from ccsnclients);

and get the error message

ERROR:  Cannot insert a duplicate key into unique index indx_unique_tbclients_ssn

 

[dmg2206]

Never mind, I've discovered the problem. The "ccsnclients" table had some duplicate values in the ssn column. I thought I had taken care of that, but apparently when we made the last update those duplicate values reappeared.

"Don't try this at home. This should only be done by trained professional idiots." :-b