PNL is a dead product

Help
2007-12-21
2013-04-16
  • Michael Wittke

    Michael Wittke - 2007-12-21

    Dear all,

    it is really sad that PNL is getting a DEAD product.
    In my opinion Intel cannot launch a product such as PNL without product improvements and an appropriate support.

    Like Imme and Gunnar wrote it in their posts, there are fundamental errors in key functions.
    And Intel does not care.

    Thus, a developer community cannot arise.

    Best regards
    Michael Wittke

    PS: I will use libDAI für my research projects: http://www.mbfys.ru.nl/~jorism/libDAI/

     
    • aharol

      aharol - 2009-07-01

      Dear all,
      I try to compile PNL on my Ubuntu 64 machine with gcc 4.3.3. Unfortunately, without success.
      I see my problem is not new, but I still cannot get in working after all the discussions here.
      The make log is
      In file included from ../../c_pgmtk/include/pnlException.hpp:33,
                       from ../../c_pgmtk/include/pnlReferenceCounter.hpp:24,
                       from ../../c_pgmtk/include/pnlMatrix.hpp:21,
                       from ../../c_pgmtk/include/pnlDenseMatrix.hpp:19,
                       from ../../c_pgmtk/src/include/pnliNumericDenseMatrix.hpp:27,
                       from ../../c_pgmtk/include/pnlNumericDenseMatrix.hpp:19,
                       from ../../c_pgmtk/include/pnlConfig.hpp:49,
                       from pnl1_5SliceInferenceEngine.cpp:17:
      ../../c_pgmtk/include/pnlString.hpp: In member function 'bool pnl::pnlString::operator==(const char*) const':
      ../../c_pgmtk/include/pnlString.hpp:68: error: 'memcmp' was not declared in this scope
      ../../c_pgmtk/include/pnlString.hpp: In member function 'bool pnl::pnlString::operator==(const std::string&) const':
      ../../c_pgmtk/include/pnlString.hpp:73: error: 'memcmp' was not declared in this scope
      ../../c_pgmtk/include/pnlString.hpp: In member function 'bool pnl::pnlString::operator==(const pnl::pnlString&) const':
      ../../c_pgmtk/include/pnlString.hpp:78: error: 'memcmp' was not declared in this scope
      ../../c_pgmtk/include/pnlString.hpp: In member function 'bool pnl::pnlString::operator<(const pnl::pnlString&) const':
      ../../c_pgmtk/include/pnlString.hpp:83: error: 'memcmp' was not declared in this scope
      make[3]: *** [pnl1_5SliceInferenceEngine.lo] Error 1
      make[3]: Leaving directory `/home/artsiom/Bayesian_nets/PNL/c_pgmtk/src'
      make[2]: *** [all-recursive] Error 1
      make[2]: Leaving directory `/home/artsiom/Bayesian_nets/PNL/c_pgmtk/src'
      make[1]: *** [all-recursive] Error 1
      make[1]: Leaving directory `/home/artsiom/Bayesian_nets/PNL/c_pgmtk'
      make: *** [all-recursive] Error 1

      Any idea would be appreciated.
      Best regards,
      Timon

       
      • Charles R. Twardy

        Looks like standard bit rot as the compilers and libraries move forward.  I can't really help yet as I never got into PNL, and I'm a research coder, not a real programmer.  (I like Python, OK?)  However, I have a GIS project coming up where licensing might not allow me to use Netica, and the GIS library is best viewed in C.

        Anyone out there interested in collaborating on a grant application to revive PNL?  I'm happy to help coordinate -- and there are enough Bayesians here at George Mason that the intersection of C and BN should be nonempty.  But I'd like to have someone experienced in PNL on board. 

        Alternatively, anyone out there want to argue this is a bad idea?

        -Charles

         
        • Jason Briceno

          Jason Briceno - 2009-07-03

          I was just getting started with PNL about a year ago for a company I was working for. I was very eager about getting it up and running again. Unfortunately, I was laid off, and I'm working somewhere else now. So, I don't have nearly enough time to contribute to the project. While I probably couldn't devote the time required to truly collaborate on a grant, you would have my full moral support (and some time) if you got started on it.

           
  • Ari Jolma

    Ari Jolma - 2010-01-28

    Add #include <cstring> to those files. Where it complains about INT_MAX not declared add #include <limits.h>.  Otherwise it should compile ok in Ubuntu 9.10.

    Ari

     
  • Alessandro Balboni

    I tried adding limits.h as suggested but during the make i receive the following error (using gcc 4.3.3):

    cvcart.cpp: In function 'BOOL cxCreatePruningStorage(CxCART*)':

    cvcart.cpp:9530: error: cast from 'CxPruningData*' to 'BOOL' loses precision

    cvcart.cpp: In function 'void cxAssertNodeValid(CxCART*, CxCARTNode*)':

    cvcart.cpp:9603: error: cast from 'CxCARTNode*' to 'unsigned int' loses precision

    cvcart.cpp:9603: error: cast from 'void*' to 'unsigned int' loses precision

    make: ***  Error 1

    make: Leaving directory `/home/balbons/Scrivania/PNL/c_pgmtk/src'

    make: ***  Error 1

    make: Leaving directory `/home/balbons/Scrivania/PNL/c_pgmtk/src'

    make: ***  Error 1

    make: Leaving directory `/home/balbons/Scrivania/PNL/c_pgmtk'

    make: ***  Error 1

    any suggestion?

     
  • Christian Rishøj

    For those interested, I have migrated the OpenPNL CVS history to GitHub, and committed a few fixes. OpenPNL now compiles and installs fine on Ubuntu 10.10.

    If anyone should like to contribute patches addressing the more central issues in the product discussed in earlier posts, please feel free to fork the project at https://github.com/crishoj/OpenPNL and send me a pull request.

    Best regards

     

Log in to post a comment.