mwgrant@adelphia.net
2003-Jul-04 04:04 UTC
[Rd] bug? exactRankTests install anomaly (PR#3394)
Full_Name: Michael W. Grant Version: 1.7.1(2003-06-16) OS: Windows 98 Second Edition 4.10.2222 A Submission from: (NULL) (24.55.165.101) I just installed version 1.7.1 tonight. In the process I installed via CRAN from the menu several packages. One installation, exactRankTests, had a distinct anomaly. A DLL attempt to change FPU control word as seem in the last line of the following listing: START LISTING:> install.packages(choose.files('',filters=Filters[c('zip','All'),]),.libPaths()[1], CRAN = NULL)> local({a <- CRAN.packages()+ install.packages(select.list(a[,1],,TRUE), .libPaths()[1], available=a)}) trying URL `http://cran.r-project.org/bin/windows/contrib/1.7/PACKAGES' Content type `text/plain; charset=iso-8859-1' length 11106 bytes opened URL downloaded 10Kb trying URL `http://cran.r-project.org/bin/windows/contrib/1.7/exactRankTests_0.8-1.zip' Content type `application/zip' length 168784 bytes opened URL downloaded 164Kb Delete downloaded files (y/N)? y updating HTML package descriptions Warning message: DLL attempted to change FPU control word from 8001f to 9001f>END LISTING NOTEs 1.) this installation was done immediately after starting Rgui, i.e., Rgui was launched, "Packages" dropdown clicked, "Install packages from CRAN..." clicked 2.) subsequently several other packages installed without incident. 3.) R was killed, restarted, and the installation done again and the same DLL/FPU anomaly occurred. Best regards
dmurdoch@pair.com
2003-Jul-04 04:45 UTC
[Rd] bug? exactRankTests install anomaly (PR#3394)
On Fri, 4 Jul 2003 04:04:13 +0200 (MET DST), you wrote:>Full_Name: Michael W. Grant >Version: 1.7.1(2003-06-16) >OS: Windows 98 Second Edition 4.10.2222 A >Submission from: (NULL) (24.55.165.101) > > >I just installed version 1.7.1 tonight. In the process I installed via CRAN from >the menu several packages. One installation, exactRankTests, had a distinct >anomaly. A DLL attempt to change FPU control word as seem in the last line of >the following listing:That is not likely a bug in R, it's a bug in some other software on your system, which has changed the floating point precision. It's hard to track down exactly what program did this, and hard to fix. The general strategy for tracking it down is to find the minimal operation that does it (e.g. does it depend on which package you install? Do you get it from a local install instead of a remote download + install, etc.) If you set options(warn.FPU=TRUE) then you'll see the warning more often; without that option, it is suppressed. Duncan Murdoch