[Rd] Should a package that indirectly Suggests: a vignette engine pass R CMD check?

[Prof Brian Ripley]

On 14/06/2014 13:54, Duncan Murdoch wrote:
> On 14/06/2014, 7:48 AM, Martin Morgan wrote:
>> A package uses VignetteEngine: knitr; the package itself does not Suggests:
>> knitr, but it Suggests: BiocStyle which in turn Suggests: knitr. Nonetheless, R
>> CMD check fails indicating that a package required for checking is not declared.
>> Is it really the intention that the original package duplicate Suggests: knitr?
>
> Yes.  The VignetteEngine setting requires that knitr be available for
> building and checking, but it doesn't imply how.  You need to also
> declare knitr in one of the other ways (Depends, Suggests, etc.).
>
> It would be possible to change the checks so that a dependency of a
> dependency was sufficient, but a suggestion of a suggestion would not
> be. You can pass a check without having the Suggests of BiocStyle
> present, because checking your package doesn't require doing anything
> with BiocStyle that would need its suggestions.  However, I don't think
> it's worth complicating the check code to support this.  If you need
> knitr to be present to build your vignettes, why not just declare it
> explicitly?

This is just checking what 'Writing R Extensions' requires:

'As Sweave is the only engine supplied with the R distribution, the 
package providing any other engine must be specified in the 
‘VignetteBuilder’ field of the package DESCRIPTION file, and also 
specified in the ‘Suggests’, ‘Imports’ or ‘Depends’ field (since its 
namespace must be available to build or check your package).'

Note the 'must be' ... it is a requirement.  And various parts of the 
software in various versions of R (there have been quite a few variants) 
have relied on this.  For example, it is optionally possible to do 
build/check without Suggests present (and always without Suggests of 
Suggests) and then in some versions some of files intended to be 
non-Sweave vignettes were silently ignored.

>
> Duncan Murdoch
>
>>
>> This is only with a recent R. In detail, with
>>
>> $ Rdev --version|head -3
>> R Under development (unstable) (2014-06-14 r65947) -- "Unsuffered Consequences"
>> Copyright (C) 2014 The R Foundation for Statistical Computing
>> Platform: x86_64-unknown-linux-gnu (64-bit)
>>
>> trying to  check the Bioconductor genefilter package leads to
>>
>> $ Rdev --vanilla CMD check genefilter_1.47.5.tar.gz
>> * using log directory ‘/home/mtmorgan/b/Rpacks/genefilter.Rcheck’
>> * using R Under development (unstable) (2014-06-13 r65941)
>> * using platform: x86_64-unknown-linux-gnu (64-bit)
>> * using session charset: UTF-8
>> * checking for file ‘genefilter/DESCRIPTION’ ... OK
>> * this is package ‘genefilter’ version ‘1.47.5’
>> * checking package namespace information ... OK
>> * checking package dependencies ... ERROR
>> VignetteBuilder package not declared: ‘knitr’
>>
>> See the information on DESCRIPTION files in the chapter ‘Creating R
>> packages’ of the ‘Writing R Extensions’ manual.
>>
>> I interpret this to mean that knitr should be mentioned in Suggests: or other
>> dependency field. The package does not Suggests: knitr, but it does Suggests:
>> BiocStyle, which itself Suggests: knitr. The author knows that they are using
>> the BiocStyle package for their vignette, and the BiocStyle package suggests the
>> appropriate builder.
>>
>> Martin Morgan
>>
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>


-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595