For SQL beginners, there's a bit of an esoteric syntax named PARTITION BY, which appears all over the place in SQL. It always has a similar meaning, though in quite different contexts. The meaning is similar to that of GROUP BY, namely to group/partition data sets by some grouping/partitioning criteria. For example, when querying the … Continue reading Various Meanings of SQL’s PARTITION BY Syntax
Category: sql
This category contains posts about SQL in general or some RDBMS’s SQL dialect
Use MULTISET Predicates to Compare Data Sets
Questions that might be a bit more difficult to solve using ordinary SQL are questions of the kind: What films have the same actors as a given film X? As always, we're using the sakila database for this example. What would be a possible way to solve this with SQL (for example, PostgreSQL, to be … Continue reading Use MULTISET Predicates to Compare Data Sets
Projecting Type Safe Nested TableRecords with jOOQ 3.17
A long standing feature request has seen little love from the jOOQ community, despite a lot of people probably wanting it. It goes by the unimpressive title Let Table<R> extend SelectField<R>: https://github.com/jOOQ/jOOQ/issues/4727 What does the feature mean, specifically? The awesome PostgreSQL Let's have a look at a really cool PostgreSQL feature. In PostgreSQL, it is … Continue reading Projecting Type Safe Nested TableRecords with jOOQ 3.17
No More MultipleBagFetchException Thanks to Multiset Nested Collections
I've recently stumbled upon this interesting Stack Overflow question about Hibernate's popular MultipleBagFetchException. The question is super popular, and the answers are plenty. The various limitations are discussed throughout the question, it all boils down to a simple fact: Joins are the wrong tool to nest collections. Given a schema like the Sakila database: There … Continue reading No More MultipleBagFetchException Thanks to Multiset Nested Collections
Approximating e With SQL
If you're running on PostgreSQL, you could try the following cool query: WITH RECURSIVE r (r, i) AS ( SELECT random(), i FROM generate_series(1, 1000000) AS t (i) ), s (ri, s, i) AS ( SELECT i, r, i FROM r UNION ALL SELECT s.ri, r.r + s.s, s.i + 1 FROM r JOIN s … Continue reading Approximating e With SQL
A Rarely Seen, but Useful SQL Feature: CORRESPONDING
I recently stumbled upon a standard SQL feature that was implemented, to my surprise, in HSQLDB. The keyword is CORRESPONDING, and it can be used with all set operations, including UNION, INTERSECT, and EXCEPT. Let's look at the sakila database. It has 3 tables with people in it: CREATE TABLE actor ( actor_id integer NOT … Continue reading A Rarely Seen, but Useful SQL Feature: CORRESPONDING
Fun with PostGIS: Mandelbrot Set, Game of Life, and More
The upcoming jOOQ 3.16 will finally offer support for the various RDBMS GIS extensions via issue #982. This is great news per se, and will be covered in a future blog post, when the integration is ready. This post here is about something else. Adding support for such a feature is a great source of … Continue reading Fun with PostGIS: Mandelbrot Set, Game of Life, and More
PostgreSQL 14’s enable_memoize For Improved Performance of Nested Loop Joins
I've recently discovered a pleasant new addition to PostgreSQL 14, the new enable_memoize flag that improves the performance of some nested loop joins where statistics hint at this being appropriate. I mean, who can resist this temptation: https://twitter.com/RPorsager/status/1455660236375826436 Improving query speed by 1000x hints at something very suboptimal having been going on before, and a … Continue reading PostgreSQL 14’s enable_memoize For Improved Performance of Nested Loop Joins
Functional Dependencies in SQL GROUP BY
The SQL standard knows an interesting feature where you can project any functional dependencies of a primary (or unique) key that is listed in the GROUP BY clause without having to add that functional dependency to the GROUP BY clause explicitly. What does this mean? Consider this simple schema: CREATE TABLE author ( id INT … Continue reading Functional Dependencies in SQL GROUP BY
The jOOQ Parser Ignore Comment Syntax
jOOQ's parser can't parse every possible SQL syntax. Try this random PostgreSQL syntax: ALTER SYSTEM RESET ALL And the jOOQ parser will complain: DOMAIN, INDEX, SCHEMA, SEQUENCE, SESSION, TABLE, TYPE, or VIEW expected: [1:7] ALTER [*]SYSTEM RESET ALL That's perfectly fine. The goal of the jOOQ parser isn't to understand all vendor specific syntax. The … Continue reading The jOOQ Parser Ignore Comment Syntax