Re: [BUGS] CC Date format code defaults to current centry - Mailing list pgsql-patches

From Karel Zak
Subject Re: [BUGS] CC Date format code defaults to current centry
Date
Msg-id 1111053051.4169.242.camel@petra
Whole thread Raw
Responses Re: [BUGS] CC Date format code defaults to current centry  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-patches
On Sun, 2005-03-13 at 21:24 -0500, Bruce Momjian wrote:
> Joe Brown wrote:
> > joe=> select to_date('19450323','CCYYMMDD');
> >   to_date
> > ------------
> >  2045-03-23
> > (1 row)
> >
> > joe=> select to_date('19450323','YYYYMMDD');
> >   to_date
> > ------------
> >  1945-03-23
> > (1 row)
> >
> > I thought the former would be "more" correct.  But it seems I am mistaken.
>
> Uh, 1945 _is_ in the 20th century, but I can see how it is confusing.

It ignored CC value if there was YY (YYYY, ..). And YY=45 is 2045:

 /*
  * 2-digit year
  */
 if (tmfc->year < 70)
   tmfc->year += 2000;
 else
   tmfc->year += 1900;


 I think it's very special case when you define YY and CC and code
should detects it and counts year as (CC-1)*100+YY. The right answers:

CC+YY

test=# select to_date('20450112', 'CCYYDDMM');
  to_date
------------
 1945-12-01

CC+Y

test=# select to_date('2090112', 'CCYDDMM');
  to_date
------------
 1909-12-01


For YYY/YYYY it ignore CC option.

The patch (with docs changes) is in the attachment.

    Karel


--
Karel Zak <zakkr@zf.jcu.cz>

Attachment

pgsql-patches by date:

Previous
From: Neil Conway
Date:
Subject: refactor preprocess_targetlist
Next
From: a_ogawa
Date:
Subject: Re: WIP: avoiding tuple construction/deconstruction overhead