Extracting time from the internet

From: George Roberts (gr@gr5.org)
Date: Fri Apr 13 2007 - 11:54:51 EDT

  • Next message: Russell Eberst: "2007APR12-13.OBS"

    I corresponded with David Herald and he has since added 
    more details including a review of Dimension 4.  It appears 
    that the code that windows uses to display the time has a bug 
    that makes it not update the display every 4th or 5th second.  
    Just double click the time and it is visually obvious.  It's not 
    clear to me if this is just a GUI bug or an internal bug with windows.
    Hopefully this is a GUI bug only.
    - George Roberts
    From: "Dave Herald" <DRHerald@bigpond.net.au>
    I've had a couple of requests to check out some other Time programs. I don't want to be
    testing a whole heap of programs - but here is a short review of the three that I've now
    looked at (all freeware).
    1. Chronos  - readily available - just do a Google Search. BUT note that it incorporates
    an optional web tracking package that you should decline to install. Fortunately the
    program asks if you want to install it - and of course you say no!
    Personally, I like the interface. You can see what is going on, and what it is checking.
    Main problem - the time that is set is around 0.5 secs slow. Noteworthy (as you will note
    later) the time given on the Chronos form clicks over at exactly the same time as the
    Windows Clock display
    2. Dimension4 - Available at http://www.thinkman.com/dimension4/download.htm
    This is apparently designed to work more behind the scenes. Unlike Chronos, it synchs to
    just one time server. You select a time server by clicking on the Server, and then OK -
    and hopefully the server responds. If the server is offline, you wouldn't notice the fact
    that no synch had occurred (at least from what I could easily see).
    Accuracy: First, the positive. The time display on the top of the Dimension4 form looks to
    be very accurate - much better than 0.1 secs. However this program obviously accesses the
    system clock at a very low level, and appears to affect the system interrupts. As a
    result, the normal windows clock display is erratic (can be slow between 0.0 and 1.0
    secs). And there were times when I had slow form refreshes.
    3. AboutTime - Available at http://www.arachnoid.com/abouttime/
    This appears to use the same protocol as Dimension4 for getting time from a server. Its
    accuracy _appears_ to be the same as Dimension4 - but there is no separate time display to
    actually check the clock accuracy. The Windows clock display is within 0.1 secs for about
    3 out of every 4 secs, and the better part of a second late for the 4th second.
    An advantage over Dimension4 is that AboutTime goes through a list of servers until it can
    synch with one - whereas in Dimension4 you apparently need to have selected a currently
    operational server. I got the sense that the impact of this package on the system
    interrupts was less than Dimension4.
    Dimension4 and AboutTime appear to give high-accuracy synchronisation to UTC - certainly
    better than 0.1secs. But the fact that the Windows clock display can be different to the
    Form display in Dimension4 is a clear indication that care is required when accessing time
    from the PC - the fact that the internal clock is accurately set does not necessarily mean
    that the displayed time is correct (at the sub-second level).
    Chronos is a more user-friendly package, but the time is late by about 0.4 secs. The
    inclusion of an optional web-monitoring package with Chronos is a distinct negative.
    Having looked at these three options, I've gone back to using Chronos - because of its
    user friendliness. But if I wanted sub-second precision, Chronos is not adequate. And
    there is a question with the other two packages of - what is the relationship between the
    time sent to (for example) CCD software and the internal clock? Given the discrepancies I
    saw between the displayed clock and the time on the Dimension4 form, there is no basis to
    _assume_ that the time sent to software is the same as the internal clock time - with
    differences of up to a second being plausible (in the absence of specific testing).
    Dave Herald
    Canberra, Australia
    Subscribe/Unsubscribe info, Frequently Asked Questions, SeeSat-L archive:  

    This archive was generated by hypermail 2b29 : Fri Apr 13 2007 - 11:56:00 EDT