Qualified wildcard to_string() is invalid about sqlparser-rs

[nickolay]

Thanks! While this fix should probably be applied anyway, I wonder how you found this.

Do you know which dialects allow qualified wildcards "inside" expressions, like count(foo.*) in your PR's testcase? ANSI SQL defines <aggregate function> ::= COUNT <left paren> <asterisk> <right paren> ..., MSSQL agrees by reporting a syntax error in count(foo.*), and the construct doesn't make sense to me. I guess COUNT(distinct foo.*) could make sense, but I don't know of a RDBMs that supports that either.

(The widely supported SELECT foo.* is handled correctly now, because we convert ASTNode::SQLQualifiedWildcard to SQLSelectItem::QualifiedWildcard here -- I was initially confused how this was not caught by an existing test.)

[thomas-jeepe]

I am not too sure exactly which databases support it, but it is a valid syntax.

Stack overflow link recommending its use.

I would have to manually test it in different databases to see which support it.

[nickolay]

Thanks! no need to test, I was wondering if you knew one off-hand. Turns out Postgres does support it, though I didn't find it mentioned in their docs.

[nickolay]

The PR was merged. Thanks again!