[R-pkg-devel] Note and warning about subdirectory sizes

Duncan Murdoch murdoch.duncan at gmail.com
Sun Oct 2 22:43:05 CEST 2016


On 02/10/2016 4:05 PM, S johnson wrote:
> Hi all,
>
> I am running check() on an R package, and receive one note and one warning
> (seem to be related):
>
> * checking installed package size ... NOTE
>   installed size is 25.9Mb
>   sub-directories of 1Mb or more:
>
>     doc       3.4Mb
>     extdata  22.1Mb
>
> * checking sizes of PDF files under ‘inst/doc’ ... WARNING
>   ‘gs+qpdf’ made some significant size reductions:
>      compacted ‘ePort.pdf’ from 1655Kb to 1076Kb
>   consider running tools::compactPDF(gs_quality = "ebook") on these files
>
> After pondering this note, I have three brief questions as follows,
>
> 1) When I run "du -hs" on the doc sub-directory, it lists its size as
> only 10K. There is only one file inside the doc sub-directory, and
> similarly, when I run "du -hs" on that one file, it lists its size as
> only 10K. I wonder, then, why check() reports that this sub-directory
> called doc is 3.4 Mb?

You are probably looking at the package source; the measurements are 
made after installation.  At that point the vignettes have been copied 
into the doc directory.

>
> 2) My extdata sub-directory is rather large. It is certainly over 1Mb.
> This is due to a few dozen example images (.png files) that we feel is
> important to include in the package. In general, do notes like these prevent
> CRAN submission acceptance? If so, what would be an alternative? I
> imagine there were other packages that had extdata subdirectory with
> example components that are larger than 1Mb - and upon a Google search
> that seems to be the case. Still, I wanted to seek your input!

There isn't a hard and fast rule.  If the data is necessary for the 
function of the package, then it will be allowed.  If it is not needed, 
then maybe not.  Often if the data is relatively stable but the code may 
be improved, you'll be asked to put them in separate packages, so that 
CRAN doesn't fill up with archived repetitions of the same data.

>
> 3) I am unfamiliar with what I should consider running compactPDF() on.
> None of the files are PDF files. The doc sub-directory has one .R file, and
> the extdata sub-directory has .R files, .png files (taking up the most
> space), and .txt files.
>

That message is likely talking about a vignette.  See the help page 
?compactPDF for how to invoke it during a build.

Duncan Murdoch



More information about the R-package-devel mailing list