Where is using a table name as a "row value" documented?

Hi,

I asked this on Stackoverflow[0] months ago, and finally posting it here as well, along with Laurenz Albe's helpful comments:

The query `SELECT * FROM table WHERE NOT (table IS NOT NULL);` finds all the rows in `table` that have a `null` value in any column. This answer[1] helped understand the logic of it, but wanted to understand this syntax. Another answer[2] states that "the reference to the table (alias) refers to an existing row" and the[PostgreSQL `SELECT` documentation's `WHERE` section[3] states that "a row satisfies the condition if it returns true when the actual **row values** are substituted for any variable references" (emphasis mine). A keyword search on "row values"[4] yielded references to the 4.2 Value Expressions[5] page, but I couldn't find an answer there either (unless I overlooked something).

> the technical term in PostgreSQL is a "whole-row reference",

> and it doesn't seem to be documented except in the source

> code. By using the table name as a column, you get a composite 

> value consisting of all columns.

>
> You can see
>
>    SELECT tab FROM tab;
>
> as being the same as
>
>    SELECT ROW(tab.*) FROM tab;

[8]: https://stackoverflow.com/a/21026085/1498178