From: Ferenc Engard <ferenc@...>
Date: 2004-01-22T05:59:05+09:00
Subject: Re: DBI improvement

"Ara.T.Howard" wrote:
> 
> On Wed, 21 Jan 2004, Ferenc Engard wrote:
> 
> > Date: Wed, 21 Jan 2004 04:43:33 +0900
> > From: Ferenc Engard <ferenc@engard.hu>
> > Newsgroups: comp.lang.ruby
> > Subject: DBI improvement
> >
> > Hello,
> >
> > I think all the DBI type classes (DBI::Timestamp etc.) should support
> > the <=> operator. Lacking this makes life heavy; implementing them would
> > be easy.
> 
> how about special timestamps?  for example, postgresql supports
> 
>   '-infinity' < all_other_times < 'infinity'
> 
> and
> 
>   'epoch' == '1970-01-01 00:00:00'
> 
> things like this get tricky.

And these can be DBI::Timestamp values? Well, then what is the result of
'-infinity'.to_time? Or 'epoch'.to_s? How gets initialized a
DBI::Timestamp object with the string 'infinity'?

I do not want to compare e.g. a Time with a DBI::Timestamp, but I want
to compare two DBI::Timestamps. Even the #== operator do something else
what one would expect.

Lacking these operators I cannot check equality of two values, or sort
the result set on client side easily, as I have to check its class, and
convert if it is DBI::*. (In fact, I can, thanks to open classes... :)

Also, the idea is a bit vague why are these types needed. For example, a
mysql timestamp is converted to DBI::Timestamp, although, if I want to
use it, I have to use as a string, as both #to_time and #to_date loses
information, and in itself, this class do not do much, even is lacks
#==!

Ferenc