abnormal exit

3 replies [Last post]
JWiley's picture
Offline
Joined: 03/25/2011

I realize I am asking for instability using development versions, but thought I would mention that running an OpenMx model kills R.

Here is my system info:

R Under development (unstable) (2012-02-03 r58258)
Platform: x86_64-pc-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=English_United States.1252
[2] LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] OpenMx_999.0.0-1937

I did an svn -update today and then in OpenMx/trunk ran: make install

Is this expected? I am a little surprised that it would build and pass those checks successfully without being able to run even the examples. The only other idea I had is that it is a permissions issue. R has permission to write/modify somethings, but depending what OpenMx calls and whether it inherits from the parent Rterm, I could see issues arising. Is this expected?

When I get a chance, I will also try building on R 2.14.1 and a more recent version of R-devel (I'm a few weeks of revisions behind).

mspiegel's picture
Offline
Joined: 07/31/2009
Can we get a little more

Can we get a little more information, such as what script are you running (generate synthetic data if necessary so we can re-create the bug) and what output is produced during the crash. Also, try running the same script in R 2.14.1 and see if it crashes.

JWiley's picture
Offline
Joined: 03/25/2011
Sorry, I thought I had

Sorry, I thought I had included the script. It is:

example(mxModel)

The build fails on R 2.14.1 (probably because the new version of R tools uses pre gcc 4.6.3 and moves its location, but here is the output from OpenMx/trunk/make install:

* building 'OpenMx_999.0.0-1937.tar.gz'

mv DESCRIPTION.bak DESCRIPTION
cd build; R CMD INSTALL "--configure-args=--enable-openmp" OpenMx_999.0.0-1937.tar.gz
* installing to library 'C:/usr/R/R-2.14.1/library'
* installing *source* package 'OpenMx' ...

** libs
Warning: this package has a non-empty 'configure.win' file,
so building only the main architecture

make[1]: Entering directory `/cygdrive/c/Users/JOSHUA~1/AppData/Local/Temp/Rtmpmu58fa/R.INSTALL137840e545d6/OpenMx/src'
x86_64-w64-mingw32-gcc -I"C:/usr/R/R-2.14.1/include" -I"d:/RCompile/CRANpkg/extralibs64/local/include" -O2 -Wall -std=gnu99 -mtune=core2 -c merge.c -o merge.o
x86_64-w64-mingw32-gcc: not found
make[1]: *** [merge.o] Error 127
make[1]: Leaving directory `/cygdrive/c/Users/JOSHUA~1/AppData/Local/Temp/Rtmpmu58fa/R.INSTALL137840e545d6/OpenMx/src'
ERROR: compilation failed for package 'OpenMx'
* removing 'C:/usr/R/R-2.14.1/library/OpenMx'
* restoring previous 'C:/usr/R/R-2.14.1/library/OpenMx'
make: *** [install] Error 1

If I just use 32-bit R 2.14.1 and use your binaries, it installs and:

example(mxModel) works fine. Anyway, this is not a high priority per se, but it may be relevant for building and using on Windows 64 bit platforms.

In terms of output produced during the crash, I am not that good at debugging on Windows, but here is the call stack leading up to it from the crash dump:

0x0
OpenMx!omxFillMatrixFromMxObjective+0x124
OpenMx!omxProcessMxAlgebraEntities+0xd4
OpenMx!callNPSOL+0x1555
R!bitwiseXor+0x5846
R!Rf_eval+0x58e
R!Rf_eval+0x1d50
R!Rf_eval+0x405
R!Rf_eval+0x1e9c
R!Rf_eval+0x405
R!Rf_applyClosure+0x367
R!Rf_eval+0x314
R!Rf_eval+0x1e9c
R!Rf_eval+0x405
R!Rf_applyClosure+0x367
R!Rf_eval+0x314
R!Rf_eval+0x1d50
R!Rf_eval+0x405
R!Rf_applyClosure+0x1e3d
R!do_mkjunction+0x6ec0
R!Rf_eval+0x113
R!Rf_applyClosure+0x367
R!do_mkjunction+0xd544
R!Rf_eval+0x113
R!Rf_applyClosure+0x367
R!do_mkjunction+0xd544
R!Rf_eval+0x113
R!Rf_applyClosure+0x367
R!Rf_eval+0x314
R!Rf_ReplIteration+0x20a
R!Rf_ReplIteration+0x572
R!run_Rmainloop+0x54
Rterm+0x16fc
Rterm+0x1584
Rterm+0x13ce
Rterm+0x14e8
kernel32!BaseThreadInitThunk+0xd
ntdll!RtlUserThreadStart+0x1d

Are people successfully building and running OpenMx on R-devel x64 on Windows? If no one is, I can try to learn more about using the Windows debugger to see if I can better trace the problem, but I'm the only one having problems, I would tend to think I screwed something up at the build stage and debugging the final crash may be an effort in futility.

mspiegel's picture
Offline
Joined: 07/31/2009
Ah, OK. We don't have a

Ah, OK. We don't have a 64-bit version of NPSOL for windows available (see FAQ). Try running the 32-bit R unstable version.