From: Paul
Date: Tue, 15 Oct 2002 12:14:43 +0100
Subject: Re: [TSSP] F A N T C: - It's NEW! - and needs some TSSP input
Kurt wrote: > Hope I'm not the only one bothering the list! Not at all! I think fantc comments should be on-list so that we can all be aware of the outstanding issues, saves duplication of effort. This application is pretty CPU intensive, and will stress the browser much more than Javascripts normally do. Marco wrote: > Each test runs on my machine in less than 10 s (_each_). That's pretty good going. Could you let us know which browser you are using? One possible outstanding issue is that the browser may defer loading the graph plotting and drawing applets until they are called for, which is at the end of the calculation. When I've downloaded it from classictesla, I've run the final demo 'Fres, C & L breakdown of bare coil' while online, just to make sure both applets are loaded. I've yet to fix this problem. The calculations will try to use 100% of the computers CPU capacity, and it's up to the operating system and browser to assign a suitable priority to this task. Too low a priority and the process will appear not to run (computer remains responsive, and fantc runs very slow). Too high priority and the program will run but everything else pretty well stops. Mouse response becomes sluggish, etc. I don't know anything about windows, but maybe there's something either in the OS or the browser options that controls the CPU priority of Javascripts and Java applets? Marco wrote: > Are we really sure the browser dowloads them just once and then > runs them always locally? Not for sure. It behaves this way on netscape/linux - I check by passing requests through a local proxy server and examine its logs. It may be setup-dependent on windows browsers. Interesting that Kurt's system refuses to run either with netscape or IE. Suggests maybe the OS is getting in the way somehow. With 256Mb ram there should be plenty of room. Kurt, if your browser won't run the geotc-test.html, then you won't get anywhere with fantc. So geotc-test is the first hurdle. I guess once you've loaded the page, your system no longer responds and you have to reboot. Therefore, can you open up the browser windows which show javascript and java error messages. On netscape/linux you get this by opening a built in page, javascript: which puts up a javascript error window, and the browser has a menu option for the java console. Hopefully, if you open these before opening geotc-test.html, then we'll see some sort of error message in one of the windows. It looks like Kurt's machine is getting hung up loading the thing in the first place. He eventually got the opening messages of the script, after 250 seconds, but there is no significant processing in the program before then. I'll make up a version of geotc-test that doesn't involve the Java applets. That will eliminate one potential source of trouble. I'll post the URL later. Stefan wrote: > one could also implement a routine to save a set of data > to the HDD. That would be a nice feature, but javascript, as far as I know, doesn't allow access to the local machine. I vaguely recall this might be overridden in Java, so maybe a Java save/restore applet is possible. This stuff is a bit out of my field and my windows/browser expert colleague is on holiday at the moment. One way to do this, avoiding the above problem, might be to get fantc to condense the system description data into a coded message (some string of characters and digits) that it can display in a message window. You then cut/paste this message into a file or document and save it yourself. Then we fit fantc with an input box into which you can paste one of these saved messages. fantc recognises it, unpacks the message into a full set of system data. That way, you can exchange 'saved systems' by pasting them directly into emails etc. So its do-able with a bit of ingenuity. -- Paul Nicholson, --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.