Skip site navigation (1) Skip section navigation (2)

Re: SET DATESTYLE to time_t style for client libraries?

From: Adam Haberlach <adam(at)newsnipple(dot)com>
To: pgsql-hackers(at)postgresql(dot)org
Subject: Re: SET DATESTYLE to time_t style for client libraries?
Date: 2002-01-03 20:32:44
Message-ID: 20020103123244.A16862@newsnipple.com (view raw or flat)
Thread:
Lists: pgsql-hackers
On Thu, Jan 03, 2002 at 03:44:50PM +0000, Thomas Lockhart wrote:
> >         So I discovered today that pgdb follows in the traditional style of
> > carrying timestamp and most other time fields through to the user as
> > text strings, so I either need to have all my queries do some gymnastics
> > to have the server format my time information in a way that is printable
> > or can be handled by my client code or whatever.
> 
> Right. Though the available styles *should* cover common usage, and
> ISO-8601 is not a bad way to go imho.

	Ok...

> > Is there a better way?  I was thinking that if there was a way to set a
> > datestyle that would just emit the seconds since the Unix epoch, I could
> > kick them into the python time module's functions for easier formatting,
> > and it would give all clients a more standardized way to deal with time
> > by letting them get the 'raw' values and handle them locally.
> 
> Hmm. If the Python module has any date/time input routines, it *should*
> be easy to ingest ISO-formatted dates. No? How about one of the other
> available styles? If nothing else, you could go through to_char() to
> format the date exactly as Python needs to see it (or directly for
> display on your client apps). date_part('epoch'...) could get you Unix
> system time, but that would last on my list...

	I'll look into getting it to ingest dates, but it seems wasteful to have
the server take its internal reprentation, pretty-format it into a nice
human-readable representation to send to the client, and then have the client
parse that into something it can deal with internally.  While it is a fairly
minor performance issue, it seems there are a lot of chances for things to
go wrong.

	I've already had to hack my python libs a bit to make the money
type work correctly.  It takes the incoming text, removes '$' and ',' and
then tries to convert it into a float.  In the case of negative values, it
will blow up because there are "()" around the value.  I'll submit a patch
if anyone is interested.

...I assume that the ISO-8601 representation itself won't be changing, but
time is silly and there's a lot of edge cases.  It'd be nice to have a way
to reliabily tell the server "Give me standardized raw values, I'll sort
things out on my end."  Of course, this may already be happening within
the C libraries and I'm not seeing them inside python.  I'll look around
a bit more.

-- 
Adam Haberlach      | Who buys an eight-processor machine and then watches 30
adam(at)newsnipple(dot)com | movies on it all at the same time?  Beats me.  They told
                    | us they could sell it, so we made it.
                    |        -- George Hoffman, Be Engineer

In response to

Responses

pgsql-hackers by date

Next:From: Tom LaneDate: 2002-01-03 20:35:31
Subject: Re: Bulkloading using COPY - ignore duplicates?
Previous:From: Hannu KrosingDate: 2002-01-03 19:55:10
Subject: Re: LWLock contention: I think I understand the problem

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group