Created attachment 338762 [details] emerge --info during a receint upgrade and complete rebuild I found that all ebuild version up to rstudio-0.97.237 fail with the following error: ========== -- Could not find libR shared library. CMake Error at /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:97 (MESSAGE): Could NOT find LibR (missing: LIBR_LIBRARIES) Call Stack (most recent call first): /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:288 (_FPHSA_FAILURE_MESSAGE) cmake/modules/FindLibR.cmake:166 (find_package_handle_standard_args) src/cpp/CMakeLists.txt:164 (find_package) ========== I will not have time to track this down in the forseeable future.
Please attach the entire build log to this bug report.
Created attachment 338952 [details] build log
+*rstudio-0.97.318 (16 Feb 2013) + + 16 Feb 2013; Julian Ospald <hasufell@gentoo.org> +rstudio-0.97.318.ebuild, + +files/rstudio-0.97.318-linker_flags.patch, + +files/rstudio-0.97.318-paths.patch, +files/rstudio-0.97.318-prefs.patch, + +files/rstudio-rserver.initd, +metadata.xml: + initial import wrt #365595 use latest in-tree version
I only see rstudio-0.97.237 in the main portage tree. Is rstudio-0.97.318 in an overlay? Or is it I just need to wait awhile to propigate? Thanks!
my bad. I as in the science overlay when I thought I was in the main tree... OK. Still no joy. I get the following error with rstudio-0.97.318: ============ -- Could not find libR shared library. CMake Error at /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:97 (MESSAGE): Could NOT find LibR (missing: LIBR_LIBRARIES) Call Stack (most recent call first): /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:288 (_FPHSA_FAILURE_MESSAGE) cmake/modules/FindLibR.cmake:166 (find_package_handle_standard_args) src/cpp/CMakeLists.txt:166 (find_package) ============ new emerge --info and build should be attached in a moment
Created attachment 339128 [details] build log for 97.318
Created attachment 339130 [details] new emerge --info for 97.318
If it is helpful I am running: dev-util/cmake-2.8.9 dev-lang/R-2.15.2-r2
(In reply to comment #5) > my bad. I as in the science overlay when I thought I was in the main tree... > > OK. Still no joy. I get the following error with rstudio-0.97.318: > > ============ > -- Could not find libR shared library. > CMake Error at > /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:97 (MESSAGE): > Could NOT find LibR (missing: LIBR_LIBRARIES) > Call Stack (most recent call first): > /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:288 > (_FPHSA_FAILURE_MESSAGE) > cmake/modules/FindLibR.cmake:166 (find_package_handle_standard_args) > src/cpp/CMakeLists.txt:166 (find_package) > ============ > > new emerge --info and build should be attached in a moment that sounds like a problem with blas can you confirm that virtual/blas is installed and that something is set via: eselect blas list
# emerge -p blas ... [ebuild R ] virtual/blas-1.0 # eselect blas list Installed BLAS for library directory lib64 [1] reference * That return value looks weird, but I would say that it is installed and selected.
(In reply to comment #10) > # emerge -p blas > ... > [ebuild R ] virtual/blas-1.0 > > # eselect blas list > Installed BLAS for library directory lib64 > [1] reference * > > That return value looks weird, but I would say that it is installed and > selected. what are your useflags for dev-lang/R? I know where the problem lies (FindLibR.cmake), but I don't know how to fix it, since I cannot reproduce.
Further to Julian's question what is the output of: R --slave --no-save -e "R.home('lib')" This should be a folder. If you get that far, we also need to know the folder's content. ls -la ${folder}
c1-srv temp # R --slave --no-save -e "R.home('lib')" WARNING: ignoring environment value of R_HOME [1] "/usr/lib64/R/lib" c1-srv temp # ls -l /usr/lib64/R/lib total 4456 -rwxr-xr-x 1 root root 1799440 Feb 15 18:35 libRlapack.so -rwxr-xr-x 1 root root 2758960 Feb 15 18:35 libR.so c1-srv temp # emerge -vp dev-lang/R ... [ebuild R ~] dev-lang/R-2.15.2-r2 USE="X bash-completion cairo java jpeg nls openmp perl png readline static-libs tiff tk -doc -icu -lapack -minimal (-prefix) -profile" 0 kB ====== I had some trouble with lapack some time back and pulled it from the system. Could that be a problem?
Could be. If not using system lapack, you get a stub from R itself. Not sure if we should just mandate lapack or not. Have to go back to the CMake rule to see what's wrong, Julian may very well do it before me.
I don't think that the lapack useflag's fault. The command I asked you to execute is what cmake uses to find the folder in which the library is located. My guess is that for some reasons it is messed up. This is rather hard to check and may require patching to do so. But first let's see all we have, can you attach /var/tmp/portage/sci-mathematics/rstudio-0.97.318/work/rstudio-0.97.318_build/CMakeCache.txt please.
Created attachment 339238 [details] CMakeCache.txt as requested Hope this is useful. Let me know if you need any more.
(In reply to comment #13) > c1-srv temp # R --slave --no-save -e "R.home('lib')" > WARNING: ignoring environment value of R_HOME > [1] "/usr/lib64/R/lib" > > c1-srv temp # ls -l /usr/lib64/R/lib > total 4456 > -rwxr-xr-x 1 root root 1799440 Feb 15 18:35 libRlapack.so > -rwxr-xr-x 1 root root 2758960 Feb 15 18:35 libR.so > > c1-srv temp # emerge -vp dev-lang/R > ... > [ebuild R ~] dev-lang/R-2.15.2-r2 USE="X bash-completion cairo java > jpeg nls openmp perl png readline static-libs tiff tk -doc -icu -lapack > -minimal (-prefix) -profile" 0 kB > > ====== > > I had some trouble with lapack some time back and pulled it from the system. > Could that be a problem? The problem is the warning message which seems to mess up the LIBR_LIB_DIR variable (which is not cached otherwise we would see it in the CMakeCache.txt). I have heard from some people solving this by reinstalling R, but please don't do that for now. What is the output for "cat /etc/env.d/99R"? Did you upgrade dev-lang/R? Did you run etc-update after that?
also do: echo $R_HOME grep ^R_HOME_DIR /usr/bin/R
# cat /etc/env.d/99R LDPATH=/usr/lib64/R/lib R_HOME=/usr/lib64/R # echo $R_HOME //usr/lib64/R # grep ^R_HOME_DIR /usr/bin/R R_HOME_DIR=/usr/lib64/R and just to see what is going on there I see: # ls -l /usr/lib64/R total 788 drwxr-xr-x 3 root root 4096 Feb 15 18:35 bin -rw-r--r-- 1 root root 18011 Feb 15 18:35 COPYING drwxr-xr-x 2 root root 4096 Feb 15 18:35 etc drwxr-xr-x 3 root root 4096 Feb 15 18:35 include drwxr-xr-x 2 root root 4096 Feb 15 18:35 lib drwxr-xr-x 82 root root 4096 Sep 21 22:12 library drwxr-xr-x 2 root root 4096 Feb 15 18:35 modules -rw-r--r-- 1 root root 249479 Feb 15 18:35 NEWS -rw-r--r-- 1 root root 501821 Feb 15 18:35 NEWS.pdf drwxr-xr-x 10 root root 4096 Sep 10 16:52 share -rw-r--r-- 1 root root 46 Feb 15 18:35 SVN-REVISION Interestingly, /usr/lib64/R/library/ was last modified last september and not during the recent update Feb 15'th. Maybe this provides a clue. ps: could we unmark this issue as FIXED?
(In reply to comment #19) > # cat /etc/env.d/99R > LDPATH=/usr/lib64/R/lib > R_HOME=/usr/lib64/R > > # echo $R_HOME > //usr/lib64/R > > # grep ^R_HOME_DIR /usr/bin/R > R_HOME_DIR=/usr/lib64/R > > and just to see what is going on there I see: > > # ls -l /usr/lib64/R > total 788 > drwxr-xr-x 3 root root 4096 Feb 15 18:35 bin > -rw-r--r-- 1 root root 18011 Feb 15 18:35 COPYING > drwxr-xr-x 2 root root 4096 Feb 15 18:35 etc > drwxr-xr-x 3 root root 4096 Feb 15 18:35 include > drwxr-xr-x 2 root root 4096 Feb 15 18:35 lib > drwxr-xr-x 82 root root 4096 Sep 21 22:12 library > drwxr-xr-x 2 root root 4096 Feb 15 18:35 modules > -rw-r--r-- 1 root root 249479 Feb 15 18:35 NEWS > -rw-r--r-- 1 root root 501821 Feb 15 18:35 NEWS.pdf > drwxr-xr-x 10 root root 4096 Sep 10 16:52 share > -rw-r--r-- 1 root root 46 Feb 15 18:35 SVN-REVISION > > Interestingly, /usr/lib64/R/library/ was last modified last september and > not during the recent update Feb 15'th. Maybe this provides a clue. > > ps: could we unmark this issue as FIXED? That is a leftover from R-2.15.2-r1 where R_HOME was incorrectly set (with two slashes instead of one) resulting in a failure in the R script which does a string comparison. Calchan has fixed that in R-2.15.2-r2, so you did not update your environment properly. R_HOME in your environment does not match R_HOME from /etc/env.d/99R fix it and recompile rstudio
I did indeed did run etc-update, but I have not rebooted my machine in over 2 weeks (I have LOTS of stuff going on in the background). The magic sause that did not happen was "source /etc/profile" after running etc-update in the root console where I do all the admin stuff. I'll see how it does now... it installed, and it comes up (I will test it properly later). Thanks, and sorry for the hastles.
