[mlpack-svn] [MLPACK] #364: remove GMM::Means() and GMM::Covariances() for 1.1.0 release

MLPACK Trac trac at coffeetalk-1.cc.gatech.edu
Wed Aug 20 15:46:10 EDT 2014


#364: remove GMM::Means() and GMM::Covariances() for 1.1.0 release
---------------------+------------------------------------------------------
 Reporter:  rcurtin  |        Owner:                                                       
     Type:  defect   |       Status:  new                                                  
 Priority:  trivial  |    Milestone:  mlpack 1.1.0                                         
Component:  mlpack   |     Keywords:  gmm, mlpack 1.1.0, means, covariances, vector, legacy
 Blocking:           |   Blocked By:                                                       
---------------------+------------------------------------------------------
 Right now, as of r16949 and r16950, the GMM class no longer uses an
 internally held vector of means and covariances.  Therefore, the `Means()`
 and `Covariances()` function issue an error if called.

 This is just fine behavior for the 'transition period' before 1.1.0, so
 that people using trunk svn can adapt their code, but we shouldn't do an
 actual 1.1.0 release where GMM holds more data members than it needs.

 Therefore I think the most reasonable solution is to keep these functions
 there until just before the 1.1.0 release, and then remove them and the
 internally held means and covariances members (which are currently there
 only so that `Means()` and `Covariances()` compiles).  Then we can clearly
 document that `Means()` and `Covariances()`Thus this ticket is just a
 reminder for whoever is working on the 1.1.0 release at release time.

 If my solution is unreasonable or could be tweaked to something better,
 feel free to tell me I am an idiot. :)

-- 
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/364>
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