Using SELECT in the WHERE clause of another SELECT

Question

I have made a draft remote application on top of libpq for PostrgreSQL. It behaves well, but I have profiled the general functioning of the application. For each final business result that I produce, it happens that I call something like 40 select clause (over tcpip).

I have reminiscences from SQL-Server reminding me to minimize the number of interactions between my remote application and the database. Having analyzed my selects, I do think I could reduce this number to 3 SELECT clauses, using joins. But I don't remember the syntax for using the result of a SELECT in another SELECT.

E.g.:

SELECT * FROM individual
INNER JOIN publisher
ON individual.individual_id = publisher.individual_id
WHERE individual.individual_id = 'here I would like to use the results of a another select'

This other SELECT would be simply of the kind:

SELECT identifier FROM another_table WHERE something='something'

Here is the simplified tables layout, declined a number of times for different item_types ... (3 totally different types, hence the 3 SQL queries if optimized).

table passage
 id_passage PK
 business_field_passage bytea
table item
 id_item PK
 id_passage FK
 business_field_item text
table item_detail
 id_item_detail PK
 id_item FK
 business_field_item_detail text
 image_content bytea

There are several id_item for one id_passage.
There are several id_item_detail for one id_item.

How would you write that?
What is the name for describing the action of redirecting one select into another (if any)?

• András Váczi: postgresql.org/docs/9.1/static/…
• Stephane Rolland: are you referring to 7.2.1.3. Subqueries ?
• András Váczi: Possibly yes, together with the JOIN part.

Answer 1

Is this what you're aiming for? Make sure the fields that are being compared are comparable (i.e. both fields are numeric, text, boolean, etc).

SELECT * FROM Individual
INNER JOIN Publisher
ON Individual.IndividualId = Publisher.IndividualId
WHERE Individual.IndividualId = (SELECT someID FROM table WHERE blahblahblah)

If you wish to select based on multiple values:

SELECT * FROM Individual
INNER JOIN Publisher
ON Individual.IndividualId = Publisher.IndividualId
WHERE Individual.IndividualId IN (SELECT someID FROM table WHERE blahblahblah)

• Stephane Rolland: can it be that straight forward ?? does it still work if SELECT someID FROM table WHERE blahblahblah has multiple records ? I'm gonna check that right now.
• Angry Spartan: Which query is selecting multiple records? It can work if you're selecting multiple records, but if you could show us your table layouts that would help us refine the answer.
• Stephane Rolland: WHERE Individual.IndividualId IN... looks good.

Answer 2

You can just rewrite that as another JOIN. This is typically simplest and fastest:

SELECT i.*, p.*
FROM individual i
JOIN publisher p USING (individualid)
JOIN another_table a ON a.identifier = i.individualid
WHERE a.something = 'something';

Only equivalent if there is at most one match in another_table. (This is true in any case if a.identifier is the PK).

I also simplified somewhat and did away with the gratuitous CaMel-case spelling of identifiers.

• Mark Storey-Smith: Yes, this. I die a little inside whenever I see IN (SELECT ..) syntax.
• Stephane Rolland: @MarkStorey-Smith Do you mean that it's more than simpler and faster: this a standard of sql coding to use another join instead of a in ( select...) In such case I should also attribute the good answer to Erwin.
• Mark Storey-Smith: @StephaneRolland Whether its faster or not will be platform and version dependent. SQL Server 2008+ for example will generate identical execution plans for INNER JOIN and IN (SELECT ...) syntax. No idea on whether same applies to PostgreSql. Performance aside, the IN (SELECT ...) style leaves me wondering if the author has fully grasped the semantics and concepts of SQL. AngrySpartan has answered your original question correctly. ErwinBrandstetter has shown you the way you should do it :).
• user1822: @MarkStorey-Smith: a JOIN is not always equivalent to an IN condition. The question is not which one is faster, the question is which one is correct.
• Clovis_Sangrail: I have only partially grasped SQL semantics/concepts. If the SELECT statement in the IN (SELECT...) portion of the overall query returns more than one row, how can the "IN (SELECT...)" syntax be replaced with JOINs ?