[mlpack-svn] [MLPACK] #304: MLpack on Scientific Linux
MLPACK Trac
trac at coffeetalk-1.cc.gatech.edu
Tue Sep 17 13:12:05 EDT 2013
#304: MLpack on Scientific Linux
--------------------+-------------------------------------------------------
Reporter: birm | Owner:
Type: defect | Status: new
Priority: minor | Milestone:
Component: mlpack | Keywords: scientific linux, RHEL6
Blocking: | Blocked By:
--------------------+-------------------------------------------------------
Essentially, this is advice to people who try to build mlpack on
scientific linux from my experiences.
the default repos do not include an up-to-date cmake, so either build from
source (using the bootstrap.sh) or add the EPEL repo to /etc/yum.repos.d,
and install cmake28.
Attempt to build normally. If that fails, recreate the build directory,
edit /usr/include/armadillo_bits/config.hpp to comment out
{{{
#define ARMA_USE_WRAPPER
}}}
and edit cmakelists.txt to include
{{{
set(ARMADILLO_LIBRARIES "${ARMADILLO_LIBRARIES};blas;lapack")
}}}
at line 61; the line after
{{{
find_package(Armadillo 2.4.2 REQUIRED)
}}}
When running cmake this time, add the argument "-D Boost_NO_BOOST_CMAKE=1"
in addition to additional arguments used.
If it builds, but tests fail due to inability to find libraries, run
"export LD_LIBRARY_PATH=/usr/local/lib"
--
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/304>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed at Georgia Tech.
More information about the mlpack-svn
mailing list