On Mon, Dec 28, 2009 at 9:20 AM, Egon Willighagen <egon.willighagen@gmail.com> wrote:
Rajarshi,

On Sat, Dec 19, 2009 at 3:45 PM, Rajarshi Guha <rajarshi.guha@gmail.com> wrote:
> On Sat, Dec 19, 2009 at 4:55 AM, Syed Asad Rahman <s9asad@googlemail.com>
> wrote:
>> While compiling the cdk code I found these warnings in abundance for
>> example
>>
>> org/openscience/cdk/qsar/model/R/PLSRegressionModelFit.java:148: warning:
>> [deprecation] org.openscience.cdk.qsar.model.R.V2M in
>> org.openscience.cdk.qsar.model.R has been deprecated
>>    [javac]         this.rmssd = V2M.VectorToMatrix(rmssd, ncase, npvar);
>>    [javac]                      ^
>
> For the warnings of this type I think it might be useful to remove the
> model.R package from master. IMO, a statistics interface is not really core
> cheminformatics. Does anybody actually use classes in this pacakge?

What are these classes doing?

They're wrappers around the JRI and R functionality, so that for example building a PLS model is simply a matter of instantiating a CDK class, rather then having to know the details of the pls() method in R.

Given that a new version of JRI and Rserve is used, which quite different from the current version used in the CDK, it'd likely require a good amount of effort to update things.

--
Rajarshi Guha
NIH Chemical Genomics Center