Re: Timestamp/Interval proposals: Part 2 - Mailing list pgsql-hackers

From Hannu Krosing
Subject Re: Timestamp/Interval proposals: Part 2
Date
Msg-id 1023719207.4092.31.camel@taru.tm.ee
Whole thread Raw
In response to Re: Timestamp/Interval proposals: Part 2  (Karel Zak <zakkr@zf.jcu.cz>)
Responses Re: Timestamp/Interval proposals: Part 2
List pgsql-hackers
On Mon, 2002-06-10 at 10:49, Karel Zak wrote:
> 
> > >  I'm _sure_ that to_char() is there for interval.
> > > 
> > > testt=# select to_char('33s 15h 10m 5month'::interval, 'HH:MI:SS Month');
> > >       to_char       
> > > --------------------
> > >  03:10:33 May
> > > (1 row)
> > 
> > Does "May" make sense for an _interval _ ? (Feb 22 + May = Jul 22)?
> > 
> > Would not "5 months" make more sense ?
> 
>  to_char() convert interval to 'tm' and make output like this struct,

My point is that to_char-ing intervals by converting them to dates is
non-intuitive.

It is really confusing to say that an interval of 5 months = "May"
and 15months == "1 March" ;(

>  I don't know what other is possible do with it.

perhaps show them with the precision specified and keep data for bigger
units in biggest specified unit.

to_char('2years 1min 4sec'::interval, 'MM SS'); ==> '24mon 64sec'
to_char('2years 1min 4sec'::interval, 'MM MI SS'); ==> '24mon 1min 4sec'


> > Or is it some ISO standard ?

Does anyone know what standard says about interval formats?

------------
annu



pgsql-hackers by date:

Previous
From: "Tony Griffiths(RA)"
Date:
Subject: Re: Missing types in C header files
Next
From: NunoACHenriques
Date:
Subject: Re: tuplesort: unexpected end of data