rnassar@duke.edu
1999-Nov-27 02:01 UTC
.Rprofile results in corruption of .RData? (PR#346)
It could be that the following may be due to something wrong that I managed to do (I cannot think what...) but I submit this report in case it is something that you want to know about (I have solved the problem by removing .Rprofile as I describe below). 1. After compiling/installing R 0.90 (an uneventful process), it worked perfectly the first time, and it read .RData written by the previous version, but later it would not read .RData (written by current version). I finally traced the problem to ~/.Rprofile which somehow seemed to cause .RData to be corrupted when R writes it, causing R to give the following message and abort next time it is run: Error: unresolved node during restore Fatal error: unable to restore saved data (remove .RData or increase memory) 2. Nothing in .Rprofile seemed suspicious. 3. I removed .Rprofile, and the problem disappeared (of course, the corrupted .RData file still could not be read, but what R now saved as .RData it could read subsequently.) 4. I copied Rprofile from /usr/local/lib/R/library/base/R to my home directory and renamed it .Rprofile without otherwise modifying it. The problem reappeared. 5. I removed ~/.Rprofile, and edited Rprofile in /usr/local/.../R (to remove significance stars and set emacs as the editor). Now everything works perfectly again (as long as there is no .Rprofile in the home directory or in my working directory. I apologize if this is some problem I caused during installation and is not a problem with R, in which case please ignore this report. I would be more than glad to try out things if you think this might be something suspicious that is happening that should be looked into. The problem on my computer is quite reproducible and does not appear to be related to the size or content of .RData. Thank you again for R! Rashid Nassar --please do not edit the information below-- Version: platform = i586-unknown-linux arch = i586 os = linux system = i586, linux status = major = 0 minor = 90.0 year = 1999 month = November day = 22 language = R Search Path: .GlobalEnv, Autoloads, Autoloads, package:base -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- 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 _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Peter Dalgaard BSA
1999-Nov-27 11:22 UTC
.Rprofile results in corruption of .RData? (PR#346)
rnassar@duke.edu writes:> It could be that the following may be due to something wrong that I > managed to do (I cannot think what...) but I submit this report in > case it is something that you want to know about (I have solved the > problem by removing .Rprofile as I describe below).We've had a similar problem report before, but it was nearly impossible to investigate because all there was to go on was the corrupted .RData (which, incidentally was readable after changing the error() in OffsetToNode to a warning(), something we should perhaps make permanent?) This could be an important lead on an elusive bug. Do you think that you could provide a minimal set of instructions that reproduce the situation? (As it stands, I cannot see whether it happens to you because of a particular call in .Rprofile, because of a specific mix of settings, or simply because .Rprofile has been called at all.) -- O__ ---- Peter Dalgaard Blegdamsvej 3 c/ /'_ --- Dept. of Biostatistics 2200 Cph. N (*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918 ~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk) FAX: (+45) 35327907 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- 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 _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._