Full_Name: Zhen Liu Version: 1.5.1 OS: Window XP Home/Window NT 4.0 Submission from: (NULL) (205.142.0.75) I have installed R-1.5.1 using SetupR.exe on both my home laptop (Sony Vaio running Window XP Home) and in office (Compaq AP550 running Window NT 4.0), but they shared the same problem. Both worked for a while, but then Rgui.exe crashed on the startup, complains: The exception Floating-point invalid operation. (0xc0000090) occurred in the application at location 0x00435c10. Despite repeated uninstalled-reboot-install, the crash still persists? Somehow R remembers it was installed on the machine before, since I can find a new machine and it will installed and work, but after a while (usually a reboot) it crashes on startup... -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-devel-request@stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
On Mon, 23 Sep 2002 17:44:42 +0200 (MET DST), you wrote in message <200209231544.RAA17579@pubhealth.ku.dk>:>Full_Name: Zhen Liu >Version: 1.5.1 >OS: Window XP Home/Window NT 4.0 >Submission from: (NULL) (205.142.0.75) > > >I have installed R-1.5.1 using SetupR.exe on both my home laptop (Sony Vaio >running Window XP Home) and in office (Compaq AP550 running Window NT 4.0), but >they shared the same problem. Both worked for a while, but then Rgui.exe crashed >on the startup, complains: > >The exception Floating-point invalid operation. >(0xc0000090) occurred in the application at location 0x00435c10. > >Despite repeated uninstalled-reboot-install, the crash still persists? Somehow R >remembers it was installed on the machine before, since I can find a new machine >and it will installed and work, but after a while (usually a reboot) it crashes >on startup... >If you start it with the command line option "--vanilla", it will skip a lot of automatically executed code at the beginning. Could you try that and see if it fixes the problem? If so, then it might be worth narrowing down which part is responsible, by adding the individual options one at a time: ‘--no-save’, ‘--no-restore’, ‘--no-site-file’, ‘--no-init-file’ and ‘--no-environ’. If --vanilla has no effect, then it might just be a coincidence that it works at first. The usual culprit for spurious floating point exceptions is code that plays around with the floating point control word, and doesn't restore it properly. Many buggy video drivers do this, for instance. I could send you some code that detects this bug, but it's hard to determine what software caused it. Duncan Murdoch -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-devel-request@stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Thanks, Duncan, for your help. I have tried each of the suggestions: C:\Program Files\R\rw1051\bin>Rgui.exe --vanilla C:\Program Files\R\rw1051\bin>Rgui.exe --no-save C:\Program Files\R\rw1051\bin>Rgui.exe --no-restore C:\Program Files\R\rw1051\bin>Rgui.exe --no-site-file C:\Program Files\R\rw1051\bin>Rgui.exe --no-init-file C:\Program Files\R\rw1051\bin>Rgui.exe --no-environ Each case results in the same crash. By the way, how do I restore the "floating point control word" ? I don't intentionally run video on my desktop, but maybe something like in real player/ms media player... Thanks again Zhen ----- Original Message ----- From: "Duncan Murdoch" <murdoch@stats.uwo.ca> To: <zhenliu66@hotmail.com> Cc: <R-bugs@biostat.ku.dk> Sent: Monday, September 23, 2002 1:01 PM Subject: Re: Rgui.exe crashes on startup (PR#2057)> On Mon, 23 Sep 2002 17:44:42 +0200 (MET DST), you wrote in message > <200209231544.RAA17579@pubhealth.ku.dk>: > > >Full_Name: Zhen Liu > >Version: 1.5.1 > >OS: Window XP Home/Window NT 4.0 > >Submission from: (NULL) (205.142.0.75) > > > > > >I have installed R-1.5.1 using SetupR.exe on both my home laptop (SonyVaio> >running Window XP Home) and in office (Compaq AP550 running Window NT4.0), but> >they shared the same problem. Both worked for a while, but then Rgui.execrashed> >on the startup, complains: > > > >The exception Floating-point invalid operation. > >(0xc0000090) occurred in the application at location 0x00435c10. > > > >Despite repeated uninstalled-reboot-install, the crash still persists?Somehow R> >remembers it was installed on the machine before, since I can find a newmachine> >and it will installed and work, but after a while (usually a reboot) itcrashes> >on startup... > > > > If you start it with the command line option "--vanilla", it will skip > a lot of automatically executed code at the beginning. Could you try > that and see if it fixes the problem? > > If so, then it might be worth narrowing down which part is > responsible, by adding the individual options one at a time: > '--no-save', '--no-restore', '--no-site-file', '--no-init-file' > and '--no-environ'. > > If --vanilla has no effect, then it might just be a coincidence that > it works at first. The usual culprit for spurious floating point > exceptions is code that plays around with the floating point control > word, and doesn't restore it properly. Many buggy video drivers do > this, for instance. I could send you some code that detects this bug, > but it's hard to determine what software caused it. > > Duncan Murdoch >-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-devel-request@stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
By the way, I have found the offending application: it is a virtual screen manager called JSP2. As soon as deinstalled it, R works just fine. Hopefully this will help other people may have similar problems. Thanks again for your help, Zhen ----- Original Message ----- From: "Duncan Murdoch" <murdoch@stats.uwo.ca> To: "Zhen Liu" <zhenliu66@hotmail.com> Sent: Tuesday, September 24, 2002 2:07 PM Subject: Re: Rgui.exe crashes on startup (PR#2057)> On Tue, 24 Sep 2002 12:49:19 -0400, you wrote in message > <OE59szTvHP0J3lTTJNJ0000205d@hotmail.com>: > > >Thanks a lot for your help, the Rgui.exe start working again!!! > > > >Here is the result when I run test87: > > > > > > > >Press test button to see effect of changing the control word. If the > >report comes without pressing the button, some driver is buggy. > > > >Try resizing this window, running other programs, etc, to see if the bug > >shows up. > > > >In R, the floating point control word should always be 037F. > >12:45:13 PM: Control word changed to 027F > >12:45:26 PM: Control word changed to 0040 > >12:46:13 PM: Control word changed to 0040 > >12:47:25 PM: Control word changed to 027F > > > >Looks like something is changing the control word to 027F, but I can't > >really figure out which program is doing that. Is there anyway to have > >Rgui.exe peg the floating point control word to what it likes? > > The difference between 37F and 27F is that 27F only gives 53 bit > precision, while 37F gives 64 bit precision. (You can see the meaning > of the bits on this web page which Google found for me: > > <http://www.compaq.com/fortran/docs/vf-html/pg/pg1contw.htm>) > > I don't think this change would cause a crash; I would have expected > to see only subtle changes in numerical results. My guess would be > that the change to 27F isn't what's causing the crash, though it might > be another symptom of the same bug. > > The difficulty in trying to fix this is that it can happen almost any > time. Whenever R makes a call to the system (e.g. to display a message > or to process a keystroke), the buggy driver might mess up the control > word. > > I'd like to try to move some code around to try to fix this, but I'm > going out of town tomorrow, and the release of 1.6.0 is imminent. > Here's what I'd suggest: > > You should try to find updates to the video drivers your system uses. > Go to the Control Panel, System, Device Manager, Display Adapters. > You probably only have one display adapter; click on it, and select > properties. I think most versions of Windows have an option to search > online for an updated driver. Try that if you can; if it can't find > an update, try the manufacturer's web site. > > After 1.6.0 is released, if you haven't been able to find a fix that > works for you, let me know and I'll play around with some changes that > might help. > > Duncan Murdoch >-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-devel-request@stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._