[mlpack-svn] [MLPACK] #206: Building mlpack with VS2010?

[MLPACK Trac]

#206: Building mlpack with VS2010?
----------------------------------+-----------------------------------------
  Reporter:  Adam                 |        Owner:     
      Type:  defect               |       Status:  new
  Priority:  major                |    Milestone:     
 Component:  mlpack               |   Resolution:     
  Keywords:  VS2010, Compilation  |     Blocking:     
Blocked By:                       |  
----------------------------------+-----------------------------------------

Comment (by rcurtin):

 Ok, thanks for the log.  It seems like a lot of the Armadillo stuff
 doesn't work nicely with the VS compiler, but I don't know how much of
 that is warnings and how much is actual errors.  It looks like all of the
 errors are in our sparse matrix extensions (in core/arma_extend/) which is
 good, at least.

 With GCC we use -Wall to help minimize the warnings in our code, but it
 looks like the Microsoft compiler is much more pedantic with -Wall.  I'll
 have to make the project work such that under Linux -Wall and -Wextra are
 used but the settings are different under Windows.

 > When using cmake it couldn't find the version of armadillo for love nor
 money, despite giving it the include directory, which for me was
 F:\Data\sdks\armadillo-2.4.3\include...I had to hard code in the version
 in the make file.

 Unfortunately CMake doesn't take either love or money, it only works using
 pure evil and crystallized sadness.  I've got a Windows laptop which I'm
 setting up now with VS2010 to try and replicate your setup -- hopefully
 then I can figure all these issues out.

 Thanks for being the first Windows guinea pig! :)  Apparently, a few years
 ago, I was told that MLPACK works just fine on Windows, but a lot has
 changed between then and now, and the guy who did that testing on Windows
 left...

-- 
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/206#comment:3>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed by the FASTLAB at Georgia Tech under Dr. Alex Gray.