Installation issue: unrecognized command line option ‘-std=c++14’

[linminhtoo]

Hi Hanjun and team,

First of all, awesome work on GLN. I am trying to pip install -e . but I keep running into this error: c++: error: unrecognized command line option ‘-std=c++14’ (full log below)
Please note that I am on an HPC cluster without root access. I think the problem might be due to the GCC version (4.8.5) being too low. I am new to Linux but I think I cannot upgrade GCC without root access. Do you happen to know any other work around for this?

Any help would be greatly appreciated. Thanks in advance and happy new year!

Note: I prepare environment this way:

TORCH_VER=1.6.0
CUDA_VER=10.1
CUDA_CODE=cu101

source /cm/shared/engaging/anaconda/2018.12/etc/profile.d/conda.sh
conda create -y -n openretro python=3.6 tqdm
conda activate openretro

conda install -y pytorch=${TORCH_VER} torchvision torchaudio cudatoolkit=${CUDA_VER} -c pytorch
conda install -y rdkit -c rdkit

# install PTG
pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-${TORCH_VER}+${CUDA_CODE}.html
pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-${TORCH_VER}+${CUDA_CODE}.html
pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-${TORCH_VER}+${CUDA_CODE}.html
pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-${TORCH_VER}+${CUDA_CODE}.html
pip install torch-geometric

# GLN installation, make sure to install on a machine with cuda
cd ./models/gln_model
pip install -e .
cd ../..

  [1/1] c++ -MMD -MF /home/linmin001/openretro/models/gln_model/build/temp.linux-x86_64-3.6/gln/mods/torchext/src/extlib.o.d -pthread -B /home/linmin001/.conda/envs/openretro/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include/torch/csrc/api/include -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include/TH -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include/THC -I/home/linmin001/.conda/envs/openretro/include/python3.6m -c -c /home/linmin001/openretro/models/gln_model/gln/mods/torchext/src/extlib.cpp -o /home/linmin001/openretro/models/gln_model/build/temp.linux-x86_64-3.6/gln/mods/torchext/src/extlib.o -fopenmp -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=extlib -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14
   FAILED: /home/linmin001/openretro/models/gln_model/build/temp.linux-x86_64-3.6/gln/mods/torchext/src/extlib.o
   c++ -MMD -MF /home/linmin001/openretro/models/gln_model/build/temp.linux-x86_64-3.6/gln/mods/torchext/src/extlib.o.d -pthread -B /home/linmin001/.conda/envs/openretro/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include/torch/csrc/api/include -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include/TH -I/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/include/THC -I/home/linmin001/.conda/envs/openretro/include/python3.6m -c -c /home/linmin001/openretro/models/gln_model/gln/mods/torchext/src/extlib.cpp -o /home/linmin001/openretro/models/gln_model/build/temp.linux-x86_64-3.6/gln/mods/torchext/src/extlib.o -fopenmp -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=extlib -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14
   c++: error: unrecognized command line option ‘-std=c++14’
   ninja: build stopped: subcommand failed.
   Traceback (most recent call last):
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/utils/cpp_extension.py", line 1515, in _run_ninja_build
       env=env)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/subprocess.py", line 438, in run
       output=stdout, stderr=stderr)
   subprocess.CalledProcessError: Command '['ninja', '-v']' returned non-zero exit status 1.
   
   During handling of the above exception, another exception occurred:
   
   Traceback (most recent call last):
     File "<string>", line 1, in <module>
     File "/home/linmin001/openretro/models/gln_model/setup.py", line 55, in <module>
       'build_ext': BuildExtension,
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/setuptools/__init__.py", line 153, in setup
       return distutils.core.setup(**attrs)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/core.py", line 148, in setup
       dist.run_commands()
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/dist.py", line 955, in run_commands
       self.run_command(cmd)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/dist.py", line 974, in run_command
       cmd_obj.run()
     File "/home/linmin001/openretro/models/gln_model/setup.py", line 45, in run
       super().run()
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/setuptools/command/develop.py", line 34, in run
       self.install_for_development()
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/setuptools/command/develop.py", line 136, in install_for_development
       self.run_command('build_ext')
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/cmd.py", line 313, in run_command
       self.distribution.run_command(command)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/dist.py", line 974, in run_command
       cmd_obj.run()
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/setuptools/command/build_ext.py", line 79, in run
       _build_ext.run(self)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/command/build_ext.py", line 339, in run
       self.build_extensions()
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/utils/cpp_extension.py", line 649, in build_extensions
       build_ext.build_extensions(self)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/command/build_ext.py", line 448, in build_extensions
       self._build_extensions_serial()
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/command/build_ext.py", line 473, in _build_extensions_serial
       self.build_extension(ext)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/setuptools/command/build_ext.py", line 196, in build_extension
       _build_ext.build_extension(self, ext)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/distutils/command/build_ext.py", line 533, in build_extension
       depends=ext.depends)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/utils/cpp_extension.py", line 478, in unix_wrap_ninja_compile
       with_cuda=with_cuda)
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/utils/cpp_extension.py", line 1233, in _write_ninja_file_and_compile_objects
       error_prefix='Error compiling objects for extension')
     File "/home/linmin001/.conda/envs/openretro/lib/python3.6/site-packages/torch/utils/cpp_extension.py", line 1529, in _run_ninja_build
       raise RuntimeError(message)
   RuntimeError: Error compiling objects for extension
   ----------------------------------------
ERROR: Command errored out with exit status 1: /home/linmin001/.conda/envs/openretro/bin/python -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/home/linmin001/openretro/models/gln_model/setup.py'"'"'; __file__='"'"'/home/linmin001/openretro/models/gln_model/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' develop --no-deps Check the logs for full command output.

[Hanjun-Dai]

Hi,

You should be able to install gcc to your local folder without ROOT (also remember to change the environment variables). Maybe the best way is to check with your HPC admin for best practice.

[vthost]

Hi! I used the above script (with same torch and cuda version) and could fix the above error so that everything seems to build fine. However, when I run the code, I get the following exception. Do you maybe have any clue where this might come from?

Installing collected packages: gln
  Running setup.py develop for gln
Successfully installed gln-0.0.0
(gln3) [vthost@cccxc536 GLN]$ cd ../
(gln3) [vthost@cccxc536 ...]$ python xyz/myglnuser.py 
Traceback (most recent call last):
  File "xyz/myglnuser.py", line 48, in <module>
    import GLN.gln.mods.mol_gnn.mg_clib.mg_lib as mg_lib
  File "/u/vthost/.../GLN/gln/mods/mol_gnn/mg_clib/__init__.py", line 1, in <module>
    from .mg_lib import MGLIB
  File "/u/vthost/.../GLN/gln/mods/mol_gnn/mg_clib/mg_lib.py", line 106, in <module>
    MGLIB = _mg_lib(sys.argv)
  File "/u/vthost/.../GLN/gln/mods/mol_gnn/mg_clib/mg_lib.py", line 14, in __init__
    self.lib = ctypes.CDLL('%s/build/dll/libmolgnn.so' % dir_path)
  File "/.../anaconda3/envs/gln3/lib/python3.6/ctypes/__init__.py", line 348, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: /u/vthost/.../GLN/gln/mods/mol_gnn/mg_clib/build/dll/libmolgnn.so: undefined symbol: _ZN3cfg14num_atom_typesE

[Hanjun-Dai]

It seems the dll is not build correctly. Can you double check it by navigating to gln/gln/mods/mol_gnn/mg_clib/ and manually do make?