[R-SIG-Mac] manually load package

Frank Harrell f.harrell at Vanderbilt.Edu
Tue Nov 22 01:52:15 CET 2011


I would rather see the most effort put into getting the maintainer of 
contrast to update.  I've been warning about rms for a year and a half.

Frank


On 11/21/2011 05:35 PM, David Winsemius wrote:
>
> On Nov 21, 2011, at 4:09 PM, Lara R. Appleby 04 wrote:
>
>> I am using the "contrast" package (available from CRAN) which in
>> turn depends on the "Design" package (no longer in CRAN).
>
> I see that you are right. It does show up in the binary packages when
> I use the GUI installer but I suspect that is because I had it before.
> From:
> Documentation for package ‘contrast’ version 0.14
> DESCRIPTION file.
> Depends: R (>= 2.2.1), Design
> Suggests: lattice, nlme, Hmisc, geepack, MASS, sandwich Description:
> Contrast methods, in the style of the Design package, for fit objects
> produced by the lm, glm, gls, and geese functions.
>> I downloaded the Design package from a web archive. How do I load
>> this into R?
>
> I was surprised to see that my older installed version of Design for R
> 2.13.1 seemed to survive the update.packages(checkBuilt=TRUE) process
> I subjected my library to. And loading 'package:contrast' does load
> Design. (Which I then needed to detach since I use 'rms' and Design
> unsurprisingly masked most of that package)
>
> If you had a prior copy of Design, you might first  try to copy it
> directly into your 2.14 library folder and then run
> update.packages("Design", checkBuilt=TRUE)
>
> If that is not an option, the Archive link at the bottom of the
> contributed packages page should have the most recent version ....
> from 2009.
>
> http://cran.r-project.org/src/contrib/Archive/Design/Design_2.3-0.tar.gz
>
>    You will need to use an Xcode-equipped version of MacOS to compile
> that source package, since it does have binaries.
>

-- 
Frank E Harrell Jr Professor and Chairman      School of Medicine
                    Department of Biostatistics Vanderbilt University



More information about the R-SIG-Mac mailing list