[Rd] Best practices - R CMD check and vignettes

[James W. MacDonald]

Thanks, Max. Right after sending that email I was frantically searching 
for the 'Retract' button on my mailer, but evidently there isn't one ;-D.

Anyway, I'm not sure the entire package is installed during either R CMD 
build or R CMD check, so I am not sure that will work. My ill-advised 
idea of hard-coding the path won't work either, since the path will be 
different for these two steps.

However, the obvious fix that occurred to me 1 nanosecond after hitting 
'send' was that the saved objects should just be in the /inst/doc 
directory with the .Rnw files instead of being in some other directory. 
Originally I had the raw data in the /examples directory so users of my 
package could practice using those data, but really I think the only use 
is for building the vignettes, so moving to /inst/doc appears the most 
reasonable course of action.

Best,

Jim



Kuhn, Max wrote:
> Jim,
> 
> But I'm not sure if this works if the package has not been installed (or
> it might find a version already installed in your library path), but
> give it a try:
> 
>    load(system.file("examples", "somedata.Rdata", package = "PkgName"))
> 
> Max 
> 
> -----Original Message-----
> From: r-devel-bounces at r-project.org
> [mailto:r-devel-bounces at r-project.org] On Behalf Of James MacDonald
> Sent: Monday, September 17, 2007 11:07 PM
> To: r-devel at stat.math.ethz.ch
> Subject: [Rd] Best practices - R CMD check and vignettes
> 
> Hi,
> 
> I have a package that contains two vignettes that both use saved objects
> 
> in the examples directory of the package. With previous versions of R I 
> could have a code chunk in the vignette like this:
> 
> <<echo=false>>=
> load("../examples/somedata.Rdata")
> @
> 
> followed by a code chunk like
> 
> <<eval=false>>=
> foo <- bar("data")
> @
> 
> that simulated the actual reading in of the data (I use a saved object 
> to limit package size).
> 
> This passed check with previous versions of R, but under R-2.6.0alpha, 
> the vignettes are dumped into an inst directory in the <package>.Rcheck 
> directory, where Sweave() and texi2dvi() are run. Unfortunately, the 
> above code chunks no longer work.
> 
> I can certainly hard code the first chunk to find the .Rdata file, but I
> 
> have to imagine there is a much more elegant way to do this.
> 
> Any suggestions?
> 
> Best,
> 
> Jim
> 
> 

-- 
James W. MacDonald, M.S.
Biostatistician
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109
734-647-5623