SQL query that will retrieve set containing all entries from another set - sql

Question

I have the following relations in my db:
Organization: information about political and economical organizations.name: the full name of the organizationabbreviation: its abbreviation
isMember: memberships in political and economical organizations.organization: the abbreviation of the organizationcountry: the code of the member country
geo_desert: geographical information about desertsdesert: the name of the desertcountry: the country code where it is locatedprovince: the province of this country
My task is to retrieve organizations which have within their members full set of countries with deserts. This organization can have also countries without deserts. So I have a set of countries with deserts and every organization in result should have all of them as members and arbitrary amount of other (no desert) countries.
I tried so far to write following code, but it doesn't work.
WITH CountriesWithDeserts AS (
SELECT DISTINCT country
FROM dbmaster.geo_desert
), OrganizationsWithAllDesertMembers AS (
SELECT organization
FROM dbmaster.isMember AS ism
WHERE (
SELECT count(*)
FROM (
SELECT *
FROM CountriesWithDeserts
EXCEPT
SELECT country
FROM dbmaster.isMember
WHERE organization = ism.organization
)
) IS NULL
), OrganizationCode AS (
SELECT name, abbreviation
FROM dbmaster.Organization
)
SELECT oc.name AS Organization
FROM OrganizationCode AS oc, OrganizationsWithAllDesertMembers AS owadm
WHERE oc.abbreviation=owadm.organization;
UPD: DBMS says: "ism.organization is not defined"
I'm using DB2/LINUXX8664 9.7.0
Output should look like this:
NAME --------------------------------------------------------------------------------
African, Caribbean, and Pacific Countries
African Development Bank
Agency for Cultural and Technical Cooperation
Andean Group

Answer 1

I find the easiest way to handle this is by using group by and having. You just want to focus on the deserts, so the rest of the countries don't matter.
select m.organization
from isMember m join
geo_desert d
on m.country = d.country
group by m.organization
having count(distinct m.country) = (select count(distinct d.country) from geo_desert);
The having clause simply counts the number of matching (i.e. desert) countries and checks that all are included.

Answer 2

Word it like this: You are looking for organizations for which not exists a desert country they don't include.
select *
from organization o
where not exists
(
select country from geo_desert
except
select country from ismember
where organization = o.abbreviation
);

Answer 3

Here are two equivalent solutions:
First:
WITH CountriesWithDeserts AS (
SELECT DISTINCT country
FROM dbmaster.geo_desert
), OrganizationsWithAllDesertMembers AS (
SELECT ism.organization
FROM dbmaster.isMember AS ism
JOIN CountriesWithDeserts AS cwd
ON ism.country = cwd.country
GROUP BY ism.organization
HAVING count(ism.country) = (SELECT count(*) FROM CountriesWithDeserts)
), OrganizationCode AS (
SELECT name, abbreviation
FROM dbmaster.Organization
)
SELECT oc.name AS Organization
FROM OrganizationCode AS oc, OrganizationsWithAllDesertMembers AS owadm
WHERE oc.abbreviation=owadm.organization;
Second:
WITH CountriesWithDeserts AS (
SELECT DISTINCT country
FROM dbmaster.geo_desert
)
SELECT org.name AS Organization
FROM dbmaster.Organization AS org
WHERE NOT EXISTS (
SELECT *
FROM CountriesWithDeserts
EXCEPT
SELECT country
FROM dbmaster.isMember
WHERE organization = org.abbreviation
);