Replied: Fri, 02 Mar 2001 17:38:05 -0500 Replied: "Ahmet Uyar" Return-Path: auyar@csit.fsu.edu Return-Path: Delivered-To: fox@csit.fsu.edu Received: from coconutpalm (unknown [144.174.137.241]) by mailer.csit.fsu.edu (Postfix) with SMTP id 8467123A01 for ; Fri, 2 Mar 2001 17:25:45 -0500 (EST) Message-ID: <00d201c0a367$bd7a6d10$f189ae90@csit.fsu.edu> From: "Ahmet Uyar" To: Subject: more time precision problems Date: Fri, 2 Mar 2001 17:25:51 -0500 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_00CF_01C0A33D.D44ECB00" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6700 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 This is a multi-part message in MIME format. ------=_NextPart_000_00CF_01C0A33D.D44ECB00 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable I discovered another time problem on PCs. The clock of my PC seems to go = a little faster, as far as I see: In every hour, my PC's clock goes 118ms faster than dirac's clock. In every hour, my PC's clock goes 136ms faster than aspen's clock. In every hour, my PC's clock goes 142ms faster than mustafa's machine's = clock. In every hour, my PC's clock goes 212ms faster than gurhan's machine's = clock. Even if we synchronize the clocks at the beginning of a session, then we = need to repeat this process maybe in every half an hour or less. Another solution can be to put the time stamps on the server side not on = the client side. First at the beginning of each session, we determine = the travel times between the server and clients. Then when server gets a = message, it subtracts the travel time of that message from the current = time, and puts the time stamp on the message. In this way, we do not = need to worry about whether client's machine clocks go faster or slower. = On the other hand, when a message takes longer than the predetermined = travel time, then the time stamp of than machine would be incorrect. Any comments? Ahmet Uyar Ph.D. candidate Computational Science and Information Technology Florida State University=20 ------=_NextPart_000_00CF_01C0A33D.D44ECB00 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

I discovered another time problem on = PCs. The clock=20 of my PC seems to go a little faster, as far as I see:

In every hour, my PC's = clock goes 118ms=20 faster than dirac's clock.

In every hour, my PC's = clock goes 136ms=20 faster than aspen's clock.

In every hour, my PC's = clock goes 142ms=20 faster than mustafa's machine's clock.

In every hour, = my PC's=20 clock goes 212ms faster than gurhan's machine's=20 clock.

Even if we synchronize the clocks at = the beginning=20 of a session, then we need to repeat this process maybe in every half an = hour or=20 less.

Another solution can be to put the time = stamps on=20 the server side not on the client side. First at the beginning = of each=20 session, we determine the travel times between the server and = clients. Then=20 when server gets a message, it subtracts the travel time of that=20 message from the current time, and puts the time stamp on the = message.=20 In this way, we do not need to worry about whether client's machine = clocks go=20 faster or slower. On the other hand, when a message takes longer than = the=20 predetermined travel time, then the time stamp of than machine would be=20 incorrect.

Any comments?

Ahmet Uyar
Ph.D. = candidate
Computational=20 Science and Information Technology
Florida State University=20

------=_NextPart_000_00CF_01C0A33D.D44ECB00--