Allow the comment prefix used in data source initialisation scripts to be configured

[szumszym]

Any line in the schema script with a sql comment and without semi-colon at the end of that line causes that the next sql statement is not executed but the rest of the script will be executed anyway.

Spring Boot v2.1.4.RELEASE

DOESNT' WORK:
DROP TABLE abc IF EXISTS;
# blablabla
DROP TABLE efg IF EXISTS;

DOES WORK:
DROP TABLE abc IF EXISTS;
# blablabla; <-- semi-colon added
DROP TABLE efg IF EXISTS;

(Please check if the issue occurs also for data.sql script)

[thackel]

I am pretty sure SQL comments are only
-- line comment
and
/* block comment */
but not starting with #.

So just change your DDL to use correct SQL comments and it will work.

[wilkinsona]

Spring Boot uses ResourceDatabasePopulator which is part of Spring Framework. By default, it uses -- for line comments and /* … */ for block comments. AIUI, these are the standard SQL comment types. Some databases, MySQL for example, also support # for line comments. The comment prefix used by ResourceDatabasePopulator can be configured, but Spring Boot does not expose this for its data source initialization. We can consider adding such a configuration option but I would recommend sticking with the standard SQL comment syntax.

[philwebb]

I think it's better that we keep ourselves aligned with the defaults in Spring Framework. A configuration option for such an edge case feels a bit too much. Only allowing standard SQL comments will also help keep scripts portable if developers ever choose to move to a different server.