PostgreSQL Query Catalog

The return value of a function cannot be left undefined. If control reaches the end of the top-level block of the function without hitting a RETURN statement, a run-time error will occur. This restriction does not apply to functions with output parameters and functions returning void, however. In those cases a RETURN statement is automatically executed if the top-level block finishes. It also does not apply to trigger functions that only task is to raise an exception.

Working with sets of rows rather than processing each row separately is more effective.

Find PL/pgSQL that contain a SELECT statement that is not a SELECT … INTO statement. This is not permitted in PL/pgSQL routines.

The value that is returned by a function should depend on something. If you do not have anything meaningful to return (for instance, the function always returns true, false, NULL, or the same number), then write VOID instead of the return type.

Find routines that have a parameter (input or output) that can have values from different types. This parameter has one of the PostgreSQL polymorphic types. The set of polymorphic types in PostgreSQL is a proper subset of its pseudo-types. The use of such a parameter allows a single routine definition to operate on many different data types, with the specific data type(s) being determined by the data types actually passed to it in a particular call.

Do not create unnecessary sequence generators.

Find columns that are not covered by a primary key, unique, and foreign key constraint but have a textual type with the maximum field size 3 or less.

Find columns of base tables that name refers to the possibility that it contains unique values but the column does not belong to any PRIMARY KEY/UNIQUE constraint. If something has to be unique, then it must be said to the system so that it could use the information for internal optimizations and enforce the constraint.

All the keys must be enforced. Find base tables that implement M:N relationship types and that allow multiple relationships of the same type between the same entities.

Find columns of base tables that contain automatically generated unique values but do not belong to any PRIMARY KEY/UNIQUE constraint. If something has to be unique, then it must be said to the system so that it could use the information for internal optimizations and enforce the constraint.

If something has to be unique, then it must be said to the system so that it could use the information for internal optimizations and enforce the constraint. Find columns that contain Universally Unique Identifiers but are not a part of any simple PRIMARY KEY/UNIQUE constraint and are also not part of a foreign key.

Find surrogate key columns that do not have an associated sequence generator. Surrogate key values must be generated by using the system (the sequence generator mechanism in case of PostgreSQL). If there are no sequence generators, then there is a question as to whether there are no surrogate keys in the database at all (could be possible and OK) or (more probable) developers have forgotten to implement the generation of surrogate keys.

Find simple natural key columns that are associated with a sequence generator.

Find sequence generators that are not associated with any column through the default value mechanism. Please note, that it is also possible to refer to a sequence generator from a routine or from an application. If these are indeed not used, then these should be dropped, otherwise these are dead code.

Find CHECK constraints that use precise comparison (= or <>) with a regular expression or LIKE pattern.

Find columns with a timestamp or a time type where the precision (the permitted maximum number of fractional sections) is bigger than the precision in the default value of the column.

Find regular expressions that do not have predefined character classes between double square brackets, e.g., [:digit:] instead of [[:digit:]].

Find routines with IF/ELSE or SWITCH/CASE statements. If your routine has a multipart IF/CASE statement, then perhaps it has multiple tasks and it violates the separation of concerns and single responsibilities principles.

Find routines with multiple raise exception commands. Perhaps it has multiple tasks and it violates the separation of concerns and single responsibilities principles.

Find for different types of database objects all the prefixes that are used in different names. One should be consistent in naming, including in the use of prefixes. If you use prefix in the name of a database object, then it could refer to the type of the database object. Do not use different prefixes in the names of database objects that have the same type.