Denormalize many-to-many

Question

I'm trying to create a denormalized view of my data that uses a typical join table to accomplish a many-to-many structure. The three tables involved are:
CREATE TABLE  "POC_SUBNETS"
   (     "ID" NUMBER,
     "OCTET1" NUMBER,
     "OCTET2" NUMBER,
     "OCTET3" NUMBER,
     "OCTET4" NUMBER,
     "SLASH" VARCHAR2(30),
     "MASK" VARCHAR2(30),
     "DESCRIPTION" VARCHAR2(255),
     "ASSIGNED" VARCHAR2(3),
      CONSTRAINT "POC_SUBNETS_PK" PRIMARY KEY ("ID") ENABLE,
      CONSTRAINT "POC_SN_ASSGN_CON" CHECK ( "ASSIGNED" in ('YES', 'NO')) ENABLE
CREATE TABLE  "POC_CONTACTS"
   (     "UNID" VARCHAR2(30),
     "FNAME" VARCHAR2(30),
     "LNAME" VARCHAR2(30),
     "DESK_PHONE" VARCHAR2(30),
     "CELL_PHONE" VARCHAR2(30),
     "PAGER" VARCHAR2(30),
     "OTHER_PHONE" VARCHAR2(30),
     "EMAIL" VARCHAR2(50),
     "ORG_ID" VARCHAR2(10),
      CONSTRAINT "POC_CONTACTS_PK" PRIMARY KEY ("UNID") ENABLE,
      CONSTRAINT "POC_CONTACTS_CON" UNIQUE ("FNAME", "LNAME") ENABLE
CREATE TABLE  "POC_SUBNET_CONTACTS"
   (     "SUBNET_ID" NUMBER,
     "UNID" VARCHAR2(30),
     "CTYPE" VARCHAR2(30),
      CONSTRAINT "POC_SUBNET_CONTACTS_PK" PRIMARY KEY ("SUBNET_ID", "UNID", "CTYPE") ENABLE
   )Obviously, this structure will allow many contacts to have many subnets but each entry in poc_subnet_contacts must be a unique combination of subnet_id, unid, and ctype.
Well, now that we've satisfied our requirement to allow this arrangement the users want to see a denormalized view of the data whereby the subnets show each of the zero to n contacts on the same row as the subnet. Something like this:
subnet_id, unid1, unid2, unid3, unidn, etc...
In truth, I can limit the number of contacts shown per subnet to 3, but I'm curious what the best way of accomplishing this denormalized view is. Here's something that I came up with:
select id, octet1, octet2, octet3, octet4,
(select unid from poc_subnet_contacts where id = subnet_id and ctype = 'LAN') LAN_UNID,
(select unid from poc_subnet_contacts where id = subnet_id and ctype = 'SEC') SEC_UNID,
(select unid from poc_subnet_contacts where id = subnet_id and ctype = 'NSC') NSC_UNID,
(select unid from poc_subnet_contacts where id = subnet_id and ctype = 'FAC') FAC_UNID
from poc_subnetsThis has the obvious limitation that the contact has to be of a specific type to appear and those may not always exist.
There also could be multiple contacts of the same type that I would like to show, if possible. This situation has caused the dreaded ORA-01427: single-row subquery returns more than one row error.
Would a materialized view help in this situation?
Earl

Answer 1

biswabijay,
Thanks for the reply. What you've suggested works similarly to my original - presumably without the problem of returning multiple rows for a single row query because of the use of MAX. That's a step in the right direction. Thank you for that.
What about the idea of being able to more dynamically add columns and the case where I might want to show multiple contacts of the same type on one row? Any ideas?
In Excel and Access this is called a crosstab query. Is there anything similar in Oracle?
Earl
P.S. I apologize in advance for mentioning the 'E' and 'A' words here on the Oracle forums. And no, I don't want to use 'E' OR 'A'. Thank you.