[mlpack-svn] [MLPACK] #357: error building mlpack
MLPACK Trac
trac at coffeetalk-1.cc.gatech.edu
Mon Aug 4 11:50:22 EDT 2014
#357: error building mlpack
-----------------------+----------------------------------------------------
Reporter: ftrovato | Owner: rcurtin
Type: defect | Status: accepted
Priority: major | Milestone:
Component: mlpack | Resolution:
Keywords: | Blocking:
Blocked By: |
-----------------------+----------------------------------------------------
Changes (by rcurtin):
* owner: => rcurtin
* status: new => accepted
Comment:
Hi Fabio,
I'm glad that you were able to solve the first issue. I was in the
process of setting up an Ubuntu 14.04 system to reproduce the issue. It
looked to me like you only had gfortran installed for i386, but you needed
it for x86_64. Either way, it seems like reinstalling the library fixed
the issue.
To solve your second issue, when you compile, add the `-lmlpack` option,
and if that still fails, tell the linker the directory that libmlpack.so
is located in with `-L/path/to/mlpack/build/lib/` (or whatever that path
may be for you, possibly just `~/mlpack-1.0.9/build/lib/`. When you run
the program, as you've correctly pointed out, you need to set
`LD_LIBRARY_PATH`.
Did libmlpack.so install into /usr/local/lib/? If so, you should be able
to just set `LD_LIBRARY_PATH=/usr/local/lib/` and that should work too.
In response to your question about the libxml2 includes, we currently
don't have a better solution to that. Some systems place libxml2 headers
in `/usr/include/` and some in `/usr/include/libxml2/`; Ubuntu is the
latter, so manually adding `/usr/include/libxml2/` to the list of include
directories is necessary. In the future I am hoping to solve this issue
by wrapping all necessary libxml2 symbols into libmlpack.so directly, but
this hasn't happened yet. Sorry for the inconvenience.
Thanks,
Ryan
--
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/357#comment:2>
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