[R-pkg-devel] Non-mainstream packages in vignette / dependency

Peter Meissner retep.meissner at gmail.com
Fri May 22 17:31:46 CEST 2015


Am .05.2015, 14:12 Uhr, schrieb Dirk Eddelbuettel <edd at debian.org>:

>
> On 22 May 2015 at 13:20, Peter Meissner via R-package-devel wrote:
> | I would like to submit a package to CRAN but I also want to use two
> | non-mainstream packages in my vignette. Currently I have them stated  
> in my
> | Suggests field but that leads to CRAN/Check complaints -understandably
> | since they cannot be obtained by install.packages().
> |
> | I know, that there is the 'Additional_repositories' field for the
> | DESCRIPTION file but as the name (and the CRAN policies) say this is  
> for
> | CRAN-like repositories and those two packages (AnomalyDetection,
> | BreakoutDetection) are on GitHub (e.g.
> | https://github.com/twitter/AnomalyDetection).
> |
> | Should I simply leave them out of the Suggests - that will lead to
> | complaints about non-stated dependencies in vignette.
> |
> | Should I fake my vignette code so that the troubling part is actually  
> not
> | run when building the vignette - that gives me headaches.
> |
> | Is there some other, clean way ?
>
> That is an excellent question that I have no answer to.
>
> It so happens that I did put both of these packages (AnomalyDetection,
> BreakoutDetection) into the 'ghrr' repository created by my drat package  
> [1].
>
> So I can install these packages fine as there is now a repo.  What I  
> have not
> done is tried to convince CRAN that such a non-CRAN, non-BioC,  
> non-Omegahat
> repo passes the implicit test set up by 'Additional_repositories'.
>
> And to the best of my knowledge, nobody else has.  So that is really a  
> CRAN
> question.  [2]
>
> Dirk
>
>
> [1] I have posted / blogged a few times about drat. It basically sits on  
> top
> of install.packages() / update.packages() by providing repositories --  
> which
> are super-easy to setup and host at GitHub via their built-in gh-pages
> webserver.  See http://dirk.eddelbuettel.com/code/drat.html as well as  
> the
> http://ghrr.github.io/drat pages for starting points, or ask me off-list.
>
> [2] So the list is zero days old and already stalled. Wow.  ;-)
>


Hey,

many thanks for the hint to your drat framework. Since it seems, that
adding 'http://ghrr.github.io/drat/' as 'Additional_repositories' in
the package DESCRIPTION file, does fulfill  the !formal! requirements,
I think I will try if I can get through with it and will report the
results to the list.


Installation works out of the box with...

install.packages(
	"AnomalyDetection",
	repos = "http://ghrr.github.io/drat",
	type="source"
)


... and first hurdle Winbuilder (devtools::build_win()) accepts the
DESCRIPTION field ...

Additional_repositories: http://ghrr.github.io/drat


... and notes the following without raising a complaint:

Suggests or Enhances not in mainstream repositories:
   AnomalyDetection, BreakoutDetection

Availability using Additional_repositories specification:
   AnomalyDetection    yes   http://ghrr.github.io/drat
   BreakoutDetection   yes   http://ghrr.github.io/drat


Now my question is what status 'http://ghrr.github.io/drat/' has. Is it a
repo bundling Github published packages open for public and going to stay
there for good? Or is it more a private thing? To put it another way.
Should I use my 'own' fork of your efforts
(http://petermeissner.github.io/drat/) or is it ok that I use ghrr and
I can assume that AnomalyDetection and BreakoutDetection stay there - I
would prefer using ghrr.


Best, Peter



-- 
Peter Meißner
Workgroup 'Comparative Parliamentary Politics'
Department of Politics and Administration
University of Konstanz
Box 216
78457 Konstanz
Germany

+49 7531 88 5665
http://www.polver.uni-konstanz.de/sieberer/home/
https://github.com/petermeissner
http://pmeissner.com



More information about the R-package-devel mailing list