The Great SQL Implementation Comparison Page

Fortunately, we have SQL standards. Or do we? It’s a well-known secret (or cynical joke) that the SQL standard is yet another SQL dialect among peers.

Standards by xkcd

On this blog, we have pointed out so many differences between SQL dialects, it is hard to believe that anyone would even consider writing SQL strings rather than using jOOQ or Hibernate if they ever risk migrating their application to another database. Just consider these examples:

• How to transform row value expressions and derived column lists
• Which databases support the row value expression BETWEEN predicate
• How to generate a range / sequence from 1 to 10 as a table
• Common, equivalent SQL function expressions

From the perspective of the jOOQ implementation, we know how tough it can be to abstract vendor-specific SQL away. Here’s an example piece of code that you don’t want to have in your application, dealing with the NVL() function:

switch (configuration.dialect().family()) {
  case ACCESS:
    return field("{iif}({0} is null, {1}, {0})");

  case DB2:
  case INGRES:
  case ORACLE:
  case H2:
  case HSQLDB:
    return field("{nvl}({0}, {1})", ..);

  case DERBY:
  case POSTGRES:
    return field("{coalesce}({0}, {1})", ..);

  case MARIADB:
  case MYSQL:
  case SQLITE:
    return field("{ifnull}({0}, {1})", ..);

  // By default, resort to the CASE expression
  default:
    return DSL.decode()
              .when(arg1.isNotNull(), arg1)
              .otherwise(arg2);
}

Now, trust us. The above is one of the easier cases. Consider the standardisation of the LIMIT .. OFFSET clause, for instance.

For an even more frightening overview, we recommend you visit Troels Arvin’s Comparison of Different SQL Implementations. Be warned, though!