Closed SionBayliss closed 3 years ago
This is a slightly tricky one, and could be caused by a CPU incompatibility, environment issue, or it may be a bug in our code.
What version of unitig-caller is this with, and does an older version work? What was the command used?
If you clone this repo and run:
python setup.py install
in your conda environment, does that fix the issue?
Hi John,
Version 1.1.0 works on both systems. The issue with version 1.2 maybe incompatibility between seqan3 and my gcc/c++ compiler in both distances (<C++17). It doesn't look to be remedied by installing gcc/c++ through conda alone (conda install gcc_linux-64 gxx_linux-64) as it still core dumps and fails to install via 'python setup.py install' with the following message:
''' CMake Error at /home/sbayliss/miniconda3/envs/test_unitigs/share/cmake/seqan3/seqan3-config.cmake:114 (message): SeqAn3 requires C++17, but your compiler does not support it. Call Stack (most recent call first): /home/sbayliss/miniconda3/envs/test_unitigs/share/cmake/seqan3/seqan3-config.cmake:237 (seqan3_config_error) CMakeLists.txt:40 (find_package) '''
This maybe due to conda installing gcc and c++ binaries with non-standard names (e.g. x86_64-conda-linux-gnu-cpp for c++). Sorry I don't have time to debug further :)
All the best, Sion
@SionBayliss I think we have fixed this in the most recent bioconda package (and 2a3da00bf67e8d503f2c0de6745caf7ab644bd4d), if you want to try updating (should have a build ending _2, but the same version number 1.2.0)
@johnlees @SionBayliss Following these instructions and then doing a conda install helped me: Please install CentOS SCL repository:
yum install centos-release-scl Install C++ support for GCC version 7:
yum install devtoolset-7-gcc-c++ --enablerepo='centos-sclo-rh' Switch enviroment to this compiler before build:
scl enable devtoolset-7 'bash'
Going to close this for now, but please reopen if still an issue
Hi John,
I am getting an 'Illegal instruction (core dumped)' error on a fresh install of unitig-caller under conda 4.9.2 (python3). I confirmed on a CLIMB instance. I have a working environment from an older conda version.
All the best, Sion