Mailing-List: contact dev-help@apr.apache.org; run by ezmlm
Precedence: bulk
Date: Fri, 4 Jan 2002 12:24:11 -0800
From: Jon Travis <jtravis@covalent.net>
To: dev@apr.apache.org
Subject: Time funniness between HPUX and everyone else
Message-ID: <20020104122411.A17981@covalent.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2.5i

I've been experiencing some time funniness on HPUX with respect
to the calculation of the GMT offset.  

Here are the results under HPUX:

    apr_explode_localtime(&xt, 0);
         xt.tm_gmtoff = -28800 
         xt.tm_isdst  = 0
 
    apr_explode_localtime(&xt, 967681226000000);
         xt.tm_gmtoff = -28800
         xt.tm_isdst  = 1


Under my Linux box:

    apr_explode_localtime(&xt, 0);
         xt.tm_gmtoff = -28800 
         xt.tm_isdst  = 0
 
    apr_explode_localtime(&xt, 967681226000000);
         xt.tm_gmtoff = -25200
         xt.tm_isdst  = 1


I believe the problem lies with the fact that we attempt to calculate
'server_gmt_offset' in the apr_unix_setup_time() function, but the gmt
offset my change when requesting times in a dst different than the
current one.  For instance, the current time is used to formulate
server_gmt_offset, and we are not in DST here in US/Pacific.  Under
Linux, we get the gmtoffset for free with our localtime call.
I thought it might be that my HPUX box was just wacked out of its
gourd wrt the timezone setup, but it looks correct after investigation.

Perhaps the correct thing to do is determine the gmt offset during dst
and also when we are not in it, and use different ones in get_offset() 
after checking dst (similar to the netware code.)

-- Jon