#66332 closed defect (fixed)

R @4.2.2_1: error when trying to install.packages possibly due to port/rpath configuration

Reported by: MarcKaufmann (Marc Kaufmann) Owned by: i0ntempest
Priority: Normal Milestone:
Component: ports Version:
Keywords: Cc: kjellpk (Kjell Konis), i0ntempest
Port: R

Description

I installed the R port, where I had to follow instructions in ticket #66095, after which it worked. The first thing I did was try to install the tidyverse package, which failed due to the package fs not installing. Here is the part of the logs where it hits the error, full logs are attached:

  CC       src/unix/libuv_la-random-getentropy.lo
  CCLD     libuv.la
ld: warning: argument missing after -rpath
ld: can't map file, errno=22 file '/opt/local/lib/libgcc' for architecture arm64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make[1]: *** [libuv.la] Error 1
make: *** [libuv-1.38.1/.libs/libuv.a] Error 2
ERROR: compilation failed for package ‘fs’

I talked with someone on the R irc who said that this looked like an issue they hit (some years ago, I think) with MacPorts installation of R. They pointed me at the following Stackoverflow question: https://stackoverflow.com/questions/26783833/rstudio-server-doesnt-find-libraries-for-macports-r-when-building

I tried to export the environment variables for CPATH and CPLUS_INCLUDE_PATH, then run an interactive R session and install the package, but nothing changed. I do not know where to set the other variables mentioned in that StackOverflow post, nor if that's what's going on.

I understand that this may not be an issue with MacPorts, but it seems to be a configuration issue with paths, so it may be. Even if not, I'd appreciate some help on where to look next or whom to ask.

Attachments (1)

R-failed-fs-installation.log (6.2 KB) - added by MarcKaufmann (Marc Kaufmann) 17 months ago.
Logs of running install.packages("fs") in an interactive R session

Download all attachments as: .zip

Change History (7)

Changed 17 months ago by MarcKaufmann (Marc Kaufmann)

Logs of running install.packages("fs") in an interactive R session

comment:1 Changed 17 months ago by kencu (Ken)

mine works fine to install "fs", but if you notice, my -Wl,-rpath entry looks slightly different than yours, and yours seems to be missing a comma:

mine:

CC="/opt/local/bin/clang-mp-14" CFLAGS="-pipe -Os -arch x86_64  -I/opt/local/include -fPIC " AR="ar" RANLIB="ranlib" LDFLAGS="-L/opt/local/lib -Wl,-headerpad_max_install_names -Wl,-rpath,/opt/local/lib/libgcc -arch x86_64" ./configure --quiet)

yours

CC="/opt/local/bin/clang-mp-14" CFLAGS="-pipe -Os -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX12.sdk -arch arm64  -I/opt/local/include -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX12.sdk -fPIC " AR="ar" RANLIB="ranlib" LDFLAGS="-L/opt/local/lib -Wl,-headerpad_max_install_names -Wl,-rpath /opt/local/lib/libgcc -Wl,-rpath,/opt/local/lib/libgcc -Wl,-syslibroot,/Library/Developer/CommandLineTools/SDKs/MacOSX12.sdk -arch arm64" ./configure --quiet)

I'm on High Sierra today.

comment:2 Changed 17 months ago by kencu (Ken)

  CCLD     libuv.la
/opt/local/bin/clang++-mp-14 -std=gnu++14 -dynamiclib -Wl,-headerpad_max_install_names -undefined dynamic_lookup -single_module -multiply_defined suppress -L/opt/local/Library/Frameworks/R.framework/Resources/lib -L/opt/local/lib -Wl,-headerpad_max_install_names -Wl,-rpath,/opt/local/lib/libgcc -arch x86_64 -o fs.so dir.o error.o file.o fs.o getmode.o id.o init.o link.o path.o utils.o unix/getmode.o ./libuv-1.38.1/.libs/libuv.a -pthread -F/opt/local/Library/Frameworks/R.framework/.. -framework R -Wl,-framework -Wl,CoreFoundation
ld: warning: text-based stub file /System/Library/Frameworks//CoreFoundation.framework/CoreFoundation.tbd and library file /System/Library/Frameworks//CoreFoundation.framework/CoreFoundation are out of sync. Falling back to library file for linking.
installing to /Users/cunningh/Library/R/x86_64/4.2/library/00LOCK-fs/00new/fs/libs
** R
** inst
** byte-compile and prepare package for lazy loading
** help
*** installing help indices
** building package indices
** installing vignettes
** testing if installed package can be loaded from temporary location
** checking absolute paths in shared objects and dynamic libraries
** testing if installed package can be loaded from final location
** testing if installed package keeps a record of temporary installation path
* DONE (fs)

comment:3 Changed 17 months ago by kencu (Ken)

now why you are missing the comma I am not currently sure about, but it appears to be in this file:

$ cat /opt/local/lib/pkgconfig/libR.pc
rhome=/opt/local/Library/Frameworks/R.framework/Versions/4.2/Resources
rlibdir=${rhome}/lib
rincludedir=/opt/local/Library/Frameworks/R.framework/Versions/4.2/Resources/include

Name: libR
Description: R as a library
Version: 4.2.2
Libs: -fopenmp  -L/opt/local/lib -Wl,-headerpad_max_install_names -Wl,-rpath,/opt/local/lib/libgcc -arch x86_64 -L${rlibdir} -lR
Cflags: -I${rincludedir} -I${rincludedir}
Libs.private:

so for a "right now" solution you might add it back there, if yours is missing it, while someone sorts out where it went.

comment:4 Changed 17 months ago by kencu (Ken)

Oh, I see it. It is here in the Portfile... this is wrong:

# See https://trac.macports.org/ticket/64204 for more info
# Remove after gcc has been fixed for arm
if {${os.arch} eq "arm"} {
    configure.ldflags-append -Wl,-rpath ${prefix}/lib/libgcc
}

and it needs to be this (or maybe removed completely):

# See https://trac.macports.org/ticket/64204 for more info
# Remove after gcc has been fixed for arm
if {${os.arch} eq "arm"} {
    configure.ldflags-append -Wl,-rpath,${prefix}/lib/libgcc
}
Last edited 17 months ago by kencu (Ken) (previous) (diff)

comment:5 Changed 17 months ago by MarcKaufmann (Marc Kaufmann)

Yes, that did the trick it seems. Although I had to edit the port file, and synching them removed the addition for issue #66095. Will I have to do that every time upon updating the package/port or will it be fixed in the next update?

Also for next time: is there something else I could have done to pinpoint the issue further? I guess you spotted it because you had a working version and could compare side by side.

Thanks for solving it so quickly, I really appreciate it.

comment:6 Changed 17 months ago by i0ntempest

Owner: set to i0ntempest
Resolution: fixed
Status: newclosed

In cd25817fe917d2b042a63e4b8d1bf5633c8d499d/macports-ports (master):

R: fix broken ldflag

Closes: #66332

Note: See TracTickets for help on using tickets.