Bug in the BuildBindings.py with Linux kernel version

Member Site Forums PyRosetta PyRosetta – Build/Install Bug in the BuildBindings.py with Linux kernel version

Viewing 2 reply threads
  • Author
    • #1959

        I think I have found a rather minor bug. It happens only with Linux. I don’t know if there is a report system for this, so I am posting it on the forum.
        Thanks for your work with Rosetta, it is really amazing.

        I am building pyrosetta from source, using the BuildPyRosetta.sh generated by rosetta/tools/PyRosetta.develop/DeployPyRosetta.py.

        During the mini-lib linking step of BuildBindings.py, a ‘change directory’ command fails because the lib_path variable is incorrect in my case:
        “cd build/src/pyrosetta/linux/3.1/64/x86/gcc/4.9/default/”
        The correct filepath is build/src/pyrosetta/linux/3.15/64/x86/gcc/4.9/default/
        lib_path’s start is set in the get_all_rosetta_objs function using :
        “if Platform == “linux”:
        lib_path = ‘build/src/’+mode+’/linux/’ + platform.release()[:3] + ‘/’ + PlatformBits +’/x86/gcc/'”
        platform.release() returns ‘3.15.5-2-ARCH’, the ‘5’ disappears.
        Conclusion: when a minor version of the Linux kernel has more than one digit, the build will fail…

        Suggested change:
        [line 793, at the beginning of get_all_rosetta_objs]
        lib_path = ‘build/src/’+mode+’/linux/’ + re.search(r'(d.d+)..*’,platform.release()).group(1) + ‘/’ + PlatformBits +’/x86/gcc/’
        This seems to solve the issue.

        Rosetta version: 2014.22.56873
        ArchLinux distribution (rolling release)
        Kernel: x86_64 Linux 3.15.5-2-ARCH
        Python 2.7.8 using virtualenv
        gcc version 4.9.1 (GCC) => unsupported, I know… Rosetta install is possible with a few tweaks. Does not matter here.

      • #10184

          Feel free to post bugs here, but you can also submit bugs on our bug tracker at http://bugs.rosettacommons.org

          I’ve submitted one for you on this issue: https://carbon.structbio.vanderbilt.edu/mantisbt/view.php?id=349

        • #10218

            Thank you for the report and path Cyril! I will apply it to mainstream.


        Viewing 2 reply threads
        • You must be logged in to reply to this topic.