Support CREATE TABLE with no columns

[benesch]

This is weird, but supported by PostgreSQL.

[coveralls]

Pull Request Test Coverage Report for Build 245

• 4 of 4 (100.0%) changed or added relevant lines in 2 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 90.102%

---

Totals
Change from base Build 234:	0.0%
Covered Lines:	3350
Relevant Lines:	3718

---

💛 - Coveralls

[nickolay]

Weird indeed and definitely not standard.

I wonder what you think of my idea about being explicit about which dialect has to support which constructs. I.e. I'd put this test into sqlparser_postgres and ran it with pg_and_generic().

To be clear: I don't ask you to move the test, rather wondering what you think about organizing the tests in such a way and if you'd object to moving it there at a later point.

[benesch]

To be clear: I don't ask you to move the test, rather wondering what you think about organizing the tests in such a way and if you'd object to moving it there at a later point.

I'm very much in favor in that! I almost wonder if we should aim for dialects to explicitly reject syntax they don't support. I.e., perhaps we should be explicitly testing that CREATE TABLE t () returns an error on all the non-Postgres dialects.