Re: T is a mandatory date time separator in RFC3339 but documentation states differently

On 2023-11-15 12:53 +0100, Peter Eisentraut wrote:
> On 15.11.23 09:37, Erik Wienhold wrote:
> > On 2023-11-15 08:16 +0100, Peter Eisentraut wrote:
> > > The SQL standard does not refer to ISO 8601 to define date formats, it has
> > > its own definitions.  In fact, PostgreSQL implements more date formats than
> > > the SQL standard requires.
> > 
> > Really?  Then what does the standard mean with section "Definitions
> > taken from ISO 8601" which I quoted in [1]?  Just using the term "date"
> > without adopting its syntax?
> 
> Exactly, it just imports the definitions of those terms.

Thanks, now I see.  SQL only defines date format 'YYYY-MM-DD' (YYYY, MM,
and DD can be any unsigned integer) with this BNF:

> <date literal> ::=
>   DATE <date string>
> 
> <date string> ::=
>   <quote> <unquoted date string> <quote>
> 
> <unquoted date string> ::=
>   <date value>
> 
> <date value> ::=
>   <years value> <minus sign> <months value> <minus sign> <days value>

And timestamp is only defined with a space separator which is clearly
not ISO 8601:

> <unquoted timestamp string> ::=
>   <unquoted date string> <space> <unquoted time string>

> > And the Postgres docs also say "The SQL standard requires the use of the
> > ISO 8601 format." [2]
> > [2] https://www.postgresql.org/docs/current/datatype-datetime.html#DATATYPE-DATETIME-OUTPUT
> 
> Yeah, that isn't correct.
> 
> I think we should reframe "ISO" to mean "ISO 9075" and remove all claims of
> alignment with ISO 8601 and RFC 3339.

Agree.  So just list the example inputs without any reference to a
particular standard, except for ISO 9075 to show that Postgres is
SQL-standard-compliant?

-- 
Erik