[R-pkg-devel] package dependencies are not available on some check severs

Duncan Murdoch murdoch.duncan at gmail.com
Wed Jan 10 18:38:05 CET 2018


On 10/01/2018 11:29 AM, Miao YU wrote:
> Hi everyone,
> 
> I want to update my package (enviGCMS) and find it can’t pass the check since some imported functions are not available.
> 
> Here is the check log: <https://win-builder.r-project.org/incoming_pretest/180110_055122_enviGCMS_050/00check.log>
> 
> Those functions are all from imported bioc packages and I have used biocViews: in DESCRIPTION . I also found current version of enviGCMS has some problems related to certain check flavor (https://CRAN.R-project.org/web/checks/check_results_enviGCMS.html).
> 
> The main issue would be that the check server just do not have the packages which is actually available on Bioc server such as xcms(http://bioconductor.org/packages/release/bioc/html/xcms.html), MSnbase(http://bioconductor.org/packages/release/bioc/html/MSnbase.html) and faahKO(https://bioconductor.org/packages/release/data/experiment/html/faahKO.html)

Generally the checks are done against R-devel, so you need to be looking 
for packages in the 3.7 version of Bioc.  This page:

http://bioconductor.org/checkResults/devel/bioc-LATEST/

shows errors in xcms and MSnbase, and there's no sign of faahKO.

> 
> As far as I could see, all those packages show a warning or error status on bioc build. Maybe the check servers on CRAN just skip those packages. Could I just import certain functions since the waining and error did not affect my imported functions?

No, if a package has errors it won't be available at all.
> 
> Any suggestions about it? Or I should wait until those imported package passed the check and then submit again?

The other choice is to find alternate sources of the functions you need. 
  If they are small and self-contained, you might be able to copy the 
source out of those packages and into yours.  If you do that, be sure to 
give proper credit to the authors.

Duncan Murdoch



More information about the R-package-devel mailing list