CREATE COLLATION - check for duplicate options and error out if found one - Mailing list pgsql-hackers

From Bharath Rupireddy
Subject CREATE COLLATION - check for duplicate options and error out if found one
Date
Msg-id CALj2ACWtL6fTLdyF4R_YkPtf1YEDb6FUoD5DGAki3rpD+sWqiA@mail.gmail.com
Whole thread Raw
Responses Re: CREATE COLLATION - check for duplicate options and error out if found one
List pgsql-hackers
Hi,

While reviewing [1], I found that the CREATE COLLATION doesn't throw an error if duplicate options are specified, see [2] for testing a few cases on HEAD. This may end up accepting some of the weird cases, see [2]. It's against other option checking code in the server where the duplicate option is detected and an error thrown if found one.  Attached a patch doing that. I chose to have the error message "option \"%s\" specified more than once" and parser_errposition because that's kind of agreed in [3].

Thoughts?

[1] https://www.postgresql.org/message-id/CALj2ACWVd%3D-E6uG5AdHD0MvHY6e4mVzkapT%3DvLDnJJseGjaJLQ%40mail.gmail.com

[2]
CREATE COLLATION coll_dup_chk (LC_COLLATE = "POSIX", LC_COLLATE = "NONSENSE", LC_CTYPE = "POSIX"); -- ERROR
CREATE COLLATION coll_dup_chk (LC_COLLATE = "NONSENSE", LC_COLLATE = "POSIX", LC_CTYPE = "POSIX"); -- OK but it's weird
CREATE COLLATION coll_dup_chk (LC_CTYPE = "POSIX", LC_CTYPE = "NONSENSE", LC_COLLATE = "POSIX"); -- ERROR
CREATE COLLATION coll_dup_chk (LC_CTYPE = "NONSENSE", LC_CTYPE = "POSIX", LC_COLLATE = "POSIX",); -- OK but it's weird
CREATE COLLATION coll_dup_chk (PROVIDER = icu, PROVIDER = NONSENSE, LC_COLLATE = "POSIX", LC_CTYPE = "POSIX"); -- ERROR
CREATE COLLATION coll_dup_chk (PROVIDER = NONSENSE, PROVIDER = icu, LC_COLLATE = "POSIX", LC_CTYPE = "POSIX"); -- OK but it's weird
CREATE COLLATION case_sensitive (LOCALE = '', LOCALE = "NONSENSE"); -- ERROR
CREATE COLLATION coll_dup_chk (LOCALE = "NONSENSE", LOCALE = ''); -- OK but it's weird
CREATE COLLATION coll_dup_chk (DETERMINISTIC = TRUE, DETERMINISTIC = NONSENSE, LOCALE = ''); -- ERROR
CREATE COLLATION coll_dup_chk (DETERMINISTIC = NONSENSE, DETERMINISTIC = TRUE, LOCALE = ''); -- OK but it's weird

[3] https://www.postgresql.org/message-id/CALj2ACUa%3DZM8QtOLPCHc7%3DWgFrx9P6-AgKQs8cmKLvNCvu7arQ%40mail.gmail.com

With Regards,
Bharath Rupireddy.
EnterpriseDB: http://www.enterprisedb.com
Attachment

pgsql-hackers by date:

Previous
From: Fujii Masao
Date:
Subject: Re: TRUNCATE on foreign table
Next
From: Aleksander Alekseev
Date:
Subject: Re: Bug fix for tab completion of ALTER TABLE ... VALIDATE CONSTRAINT ...