As I'm working my way through Celkos "SQL for Smarties", he points out an interesting way to map such boolean queries by using GROUP BY and HAVING:

-- find all book ids coauthored by Smith and Jones:
-- (let's assume that the author_name is unique)
select book_id, count(*) from authors
where 
  author_name = 'Smith'
  or author_name = 'Jones'
group by book_id
having count(*) = 2
[download]

Whether the performance of this is better or worse than selecting all books by one and all books by the other and scanning the lists in Perl, or worse than dynamically creating the self joins needs to be benchmarked. Another way to make this slightly less dynamic could be to use the IN predicate:

-- find all book ids coauthored by Smith and Jones:
-- (let's assume that the author_name is unique)
select book_id, count(*) from authors
where 
  author_name in ('Smith', 'Jones')
group by book_id
having count(*) = 2
[download]

To make it completely static, you could insert the authors into a (session-local) temporary table and cross-join that table.

Update: Oops - this approach will also return books co-authored by 'Smith', 'Jones' and 'Miller'. Depending on whether that's what you want or not, you will need to use a subselect to make sure you get all of the authors. But I'm not sure if I can come up with the correct subselect at the moment :-)

Well, you can fix it by using except on a query that gives all the book_id's where the author isn't one of 'Smith' or 'Jones'.

select book_id from authors
where
  author_name in ('Smith','Jones')
group by book_id
having count(*)=2
except
select book_id from authors
where
  author_name not in ('Smith','Jones');
[download]

You could also alter the where clause to make certain book_id doesn't have a different author:

select book_id from authors
where
  author_name in ('Smith','Jones')
  and book_id not in (
    select book_id from authors
    where
       author_name not in ('Smith','Jones')
  )
group by book_id
having count(*)=2
[download]

By the way, if anyone wants to read some of Joe Celko's old articles from DBMS magazine, here are some that are available from their website (the puzzles are really fun).

This is really off-topic for this site, but let's have a go anyway.

You need to join the BookAuthor table with itself. I'm old-skool so I do joins in the WHERE clause, but there's probably a simpler way to do it using an explicit JOIN clause.

select a.book
from BookAuthor a, BookAuthor b
where a.author = 'Smith' and b.author = 'Jones'
and a.book = b.book;
[download]

Update: The JOIN syntax looks like this:

select a.book
from BookAuthor a
inner join BookAuthor b
on a.book = b.book
where a.author = 'Smith' and b.author = 'Jones';
[download]

You should be able to omit the "INNER", but I was testing on a rather old version of MySQL which didn't support the abbreviated syntax.

SELECT book FROM BooksAuthors WHERE Author in ('Jones', 'Smith');
[download]

Also look at SQL::Abstract and the extension to Class::DBI that uses it, Class::DBI::AbstractSearch, for an easy way to build these queries dynamically using perl data structures, for any length list of authors.

That's close but will also return books authored by only one of 'Jones' or 'Smith' alone. Of course, some quick Perl filtering might be easier than doing it in SQL.

I don't think that's true, unless different SQL engines implement IN differently.
On all the engines I've used, IN is equivalent to a list of OR statements.

Just convert everything to upper case. Far more efficient than using LIKE unnecessarily.

SELECT book
FROM   BooksAuthors AS B1
JOIN   BooksAuthors AS B2
ON     B1.Book = B2.Book
WHERE  UPPER(B1.Author) = 'JONES'
AND    UPPER(B2.Author) = 'SMITH'
[download]

--
<http://dave.org.uk>

"The first rule of Perl club is you do not talk about Perl club."
-- Chip Salzenberg