Обсуждение: Functions vs. Columns?????

Hi,

I have a question as to how postgres handles some built-in functions,
for example:

Works                        Doesn't Work
------                       ------------
select now();                select now;
select current_date;         select current_date();

It seems that there should be some consistency here in the use of
parenthesis.  I don't really know how these things are handled on the
backend, but it seems that parenthesis are the more correct way, or at
least both should be allowed.

Any thoughts?

Byron

> -----Original Message-----
> From: owner-pgsql-interfaces@postgreSQL.org
> [mailto:owner-pgsql-interfaces@postgreSQL.org]On Behalf Of Byron
> Nikolaidis
> 
> Hi,
> 
> I have a question as to how postgres handles some built-in functions,
> for example:
> 
> Works                        Doesn't Work
> ------                       ------------
> select now();                select now;
> select current_date;         select current_date();
> 
> It seems that there should be some consistency here in the use of
> parenthesis.  I don't really know how these things are handled on the
> backend, but it seems that parenthesis are the more correct way, or at
> least both should be allowed.
> 
> Any thoughts?
>

Parenthesis are needed in order to call functions in PostgreSQL.
current_date isn't a built-in function of PostgreSQL.
It's an SQL standard function and it's a special work of PostgreSQL
parser to change current_date to call another built-in function.
Regards.

Hiroshi Inoue
Inoue@tpf.co.jp

Byron Nikolaidis <byron.nikolaidis@home.com> writes:
> Works                        Doesn't Work
> ------                       ------------
> select now();                select now;
> select current_date;         select current_date();

> It seems that there should be some consistency here in the use of
> parenthesis.

Consistency?  Who needs consistency ;-) ?

What you're seeing here is the conflict of two cultures.  The SQL92
standard mandates a keyword CURRENT_DATE, with *no* parentheses,
as the way to get the current date.  now() comes out of the Berkeley
Postquel code, which inherits from an academic tradition that doesn't
like inventing reserved words without need, and also thinks that
things that look like constants ought to *be* constants.  The way
to resolve that tension, in academese, is to invent parameterless
functions.  But the folks who wrote the SQL92 spec seem to have been
raised on COBOL, which never met a reserved word it didn't like.

Anyway, Postgres supports both the SQL92 notation and the notation
we inherited from Postquel.  I don't think that adding with-or-
without-parentheses-take-your-pick-for-both syntax freedom would be
a step forward... that path leads to way too many traps for the
unwary programmer.
        regards, tom lane

> It seems that there should be some consistency here in the use of
> parenthesis.  I don't really know how these things are handled on the
> backend, but it seems that parenthesis are the more correct way, or at
> least both should be allowed.

The ones without parens are SQL-mandated "constants" which are
translated in the parser front end into function calls (usually with
some other name for historical reasons). Other function calls, like
now(), are really that: function calls.

> Any thoughts?

Uh, sure ;)
                    - Thomas

-- 
Thomas Lockhart                lockhart@alumni.caltech.edu
South Pasadena, California