在postgresql中，函数有个不稳定性分类属性，它会影响优化器评估函数的可优化级别、同时也会影响postgresql中并行执行的可行性。
函数有三种类型：VOLATILE、STABLE以及IMMUTABLE。VOLATILE是函数默认类别，也就是优化器假设函数会修改数据库，不会做任何特定的优化。STABLE可用于基于函数的索引扫描。IMMUTABLE通常用于静态常量的优化。对于一些短小逻辑的函数，典型的是数据字典翻译，但是写在SQL中会导致很复杂的逻辑，应该在创建函数时声明为STABLE。一般来说IMMUTABLE不是特别必要，更像是resultset_cache的实现。如下：
lightdb@postgres=# CREATE INDEX idx1_lem_db_log ON lem_db_log USING gin(to_tsquery(concat('field1='|| field1,' field2=' || field2,' field3=' || field3,' field4=' || field4,' field5='||field5)));
ERROR: functions in index expression must be marked IMMUTABLE
可见在函数索引中，表达式必须为不可变的。要修改系统函数的函数属性，可通过alter table函数签名进行修改，如下：
lightdb@postgres=# \df+ concat()
List of functions
Schema | Name | Result data type | Argument data types | Type | Volatility | Parallel | Owner | Security | Access privileges | Language | Source code | Description
------------+--------+------------------+---------------------+------+------------+----------+---------+----------+-------------------+----------+-------------+--------------------
pg_catalog | concat | text | VARIADIC "any" | func | stable | safe | lightdb | invoker | | internal | text_concat | concatenate values
(1 row)
lightdb@postgres=# alter function pg_catalog.concat(variadic "any") immutable; -- 任何长度、任意类型的参数
ALTER FUNCTION
lightdb@postgres=# CREATE INDEX idx1_lem_db_log ON lem_db_log USING gin(to_tsquery(concat('field1='|| field1,' field2=' || field2,' field3=' || field3,' field4=' || field4,' field5='||field5)));
ERROR: functions in index expression must be marked IMMUTABLE
lightdb@postgres=# \df+ to_tsvector()
List of functions
Schema | Name | Result data type | Argument data types | Type | Volatility | Parallel | Owner | Security | Access privileges | Language | Source code |
Description
------------+-------------+------------------+---------------------+------+------------+----------+---------+----------+-------------------+----------+-------------------------------+--------
----------------------------------------
pg_catalog | to_tsvector | tsvector | json | func | stable | safe | lightdb | invoker | | internal | json_string_to_tsvector | transfo
rm string values from json to tsvector
pg_catalog | to_tsvector | tsvector | jsonb | func | stable | safe | lightdb | invoker | | internal | jsonb_string_to_tsvector | transfo
rm string values from jsonb to tsvector
pg_catalog | to_tsvector | tsvector | regconfig, json | func | immutable | safe | lightdb | invoker | | internal | json_string_to_tsvector_byid | transfo
rm string values from json to tsvector
pg_catalog | to_tsvector | tsvector | regconfig, jsonb | func | immutable | safe | lightdb | invoker | | internal | jsonb_string_to_tsvector_byid | transfo
rm string values from jsonb to tsvector
pg_catalog | to_tsvector | tsvector | regconfig, text | func | immutable | safe | lightdb | invoker | | internal | to_tsvector_byid | transfo
rm to tsvector
pg_catalog | to_tsvector | tsvector | text | func | stable | safe | lightdb | invoker | | internal | to_tsvector | transfo
rm to tsvector
(6 rows)
多个重载的时候，要找相同的那个重载实现。修改之后，函数索引即可创建，如下：
lightdb@postgres=# CREATE INDEX idx1_lem_db_log ON lem_db_log USING gin(to_tsvector('simple',concat('field1='|| field1,' field2=' || field2,' field3=' || field3,' field4=' || field4,' field5='||field5)));
CREATE INDEX
CREATE FUNCTION dup(in int, out f1 int, out f2 text)
AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
LANGUAGE sql IMMUTABLE ;
explain analyze SELECT dup(42) from (select 1 from pg_catalog.generate_series(1, 1000000));
http://www.light-pg.com/docs/lightdb/13.3-22.2/xfunc-volatility.html
http://www.light-pg.com/docs/lightdb/13.3-22.2/sql-createfunction.html
http://www.light-pg.com/docs/lightdb/13.3-22.2/xfunc-pl.html
