wb@arb-phys.uni-dortmund.de
2004-Jun-22 20:53 UTC
[Rd] lme4 fails to install on R-1.9/FreeBSD-5.2 (PR#7007)
Full_Name: W.B.Kloke Version: 1.9.1 OS: FreeBSD-5.2.1 Submission from: (NULL) (195.253.16.182) Subject line says it. I had problems installing lme4. 1. The dependency on package Matrix was not resolved (I am not sure that this is really a bug; but it is annoying, anyway). 2. Installing Matrix failed with a message saying something like "no rule for %_D.o" after compiling a lot of sources and combining them into an archive. I traced this down to a Linuxism. I found the string _D.o in subdirectory taucs of Matrix. Forcing GNU make by adding MAKE=gmake to the environment of R installed the package successfully. I consider this a real non-portability bug.
Prof Brian Ripley
2004-Jun-23 14:30 UTC
[Rd] lme4 fails to install on R-1.9/FreeBSD-5.2 (PR#7007)
Please send reports on contributed packages to the package maintainer (Cc:ed here, but he is on leave this summer so don't expect a fast response). On Tue, 22 Jun 2004 wb@arb-phys.uni-dortmund.de wrote:> Full_Name: W.B.Kloke > Version: 1.9.1What about the package versions?> OS: FreeBSD-5.2.1 > Submission from: (NULL) (195.253.16.182) > > > Subject line says it. I had problems installing lme4.Actually not (especially as there is no R-1.9). I gather you had problems installing Matrix, and the bug you found is in Matrix.> 1. The dependency on package Matrix was not resolved (I am not sure that this is > really a bug; but it is annoying, anyway).I am not sure what you mean here. It seems you did not install Matrix first? Did you expect install.packages() to do that for you?> 2. Installing Matrix failed with a message saying something like "no rule for > %_D.o" > after compiling a lot of sources and combining them into an archive. I traced > this down to a Linuxism. I found the string _D.o in subdirectory taucs of > Matrix. Forcing GNU make by adding MAKE=gmake to the environment of R installed > the package successfully. I consider this a real non-portability bug.GNUism, I suspect. There seem to be similar problems on Solaris. make: Fatal error in reader: Makefile, line 32: Extra `:', `::', or `:=' on dependency line Current working directory /tmp/R.INSTALL.28676/Matrix/src/taucs I managed to get this to compile and check (apart from some documentation problems) with %_D.o: %.c $(CC) $(ALL_CPPFLAGS) $(ALL_CFLAGS) $(TDFLAGS) -c $< -o $@ %.o: %.c $(CC) $(ALL_CPPFLAGS) $(ALL_CFLAGS) $(TGFLAGS) -c $< -o $@ (there is no TAUCS_G to worry about). I am sure Dr Bates will appreciate your sending him a tested patch when you get it working. -- Brian D. Ripley, ripley@stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595
Possibly Parallel Threads
- problem building R 2.9.1 from source on RHEL 4 (x86_64)
- using so-library involving Taucs
- [LLVMdev] Including svn version number in --version output
- Developmental version of Matrix package for R-1.9.0
- [LLVMdev] Including svn version number in --version output