On 22 Mar, Bob Latham wrote in message <54a881b521bob@xxxxxxxxxxxxxxxx>: > In article <mpro.nlmju90901fn302em.lists@xxxxxxxxxxxxxxxxxx>, > Steve Fryatt <lists@xxxxxxxxxxxxxxxxxx> wrote: > > On 22 Mar, Bob Latham wrote in message > > <54a8639c96bob@xxxxxxxxxxxxxxxx>: > > > > > I then installed a copy on my wife's ARMiniX where the clock was well > > > over 3 minutes slow which I was completely unable to correct with the > > > OS functionality. Your app set my machine and my wife's with a second > > > of each other which is good enough for me. > > > What was NetTime reporting it thought was happening at the time? Did it > > think the time was correct, or was it in the process of adjusting it? > > Sorry Steve, That question needs a deeper understanding that I have. All I > saw was the time out by more than 3 minutes and "busy" in the little box. > I tried the "try" button but if I'm honest I was only guessing what might > do but it didn't help and I've no idea what in particular was busy. I don't really know what the GUI does, but "Busy" sounds as if it's in the process of adjusting the time for you. At the command-line (Ctrl-F12), you can do *nettime_status to ask the module what's going on. Trying it here just now gives: *nettime_status Current time: Sunday, 22 March 2015 22:53:50.33 Status: Sleeping Last adjustment: 12 minutes 37 seconds ago Last delta: 0.000918 seconds slow Last server: blencathra Last protocol: SNTP Poll interval: 15 minutes Timer latch: 130000 * which tells me that the clock on this machine is pretty close to what the server is reporting (blencatha is a local Linux box that syncs to uk.pool.ntp.org and then offers NTP services to the RISC OS machine, so that only one machine is asking for the time over the internet). > > If you think you've found a problem, details of NetTime's status and a > > bug report to ROOL would be useful. > > I was surprised to find I could find a way to get the machine to fetch the > time and adjust the clock. It may well be that it's not designed to do > that. It isn't designed to do that, unless the delta is much bigger. AFAIK no modern system (Windows, Linux, etc) will do it for small changes, because stepping back in time can cause problems for anything that relies on reading chronological file datestamps. > > > I would like to see something added to the OS for this, 3 minutes > > > adrift and not corrected is a bit poor to say the least. > > > NetTime does not, for very good reasons, force a step change on the > > clock for relatively small errors. Instead, it speeds up or slows down > > the clock until it converges with reality (and then continues to adjust > > the length of a tick so that the two stay in step thereafter). > > Yes, I can see that now but it is quite frustrating not to be able to set > the clock by my choice. I don't think I've ever encountered a clock that > had that feature before in my life. Good luck even finding a configuration option on a modern OS (you have to hunt around on the commandline to even change the NTP server on Ubuntu Linux, for example). > > This is the same approach that other systems use, and it avoids problems > > such as time apparently going backwards as far as applications are > > concerned. > > Yes, I'm sure you're right but I've never come across it before and to be > honest, I don't like the restriction. NetTime's worked in the same way since the early noughties without anyone complaining, and it does nothing different to what RTCAdjust has done since the dawn of the nineties. -- Steve Fryatt - Leeds, England http://www.stevefryatt.org.uk/ --- To alter your preferences or leave the group, visit //www.freelists.org/list/armini-support List-related queries to info@xxxxxxxxxxxx