[BioC] R CMD check : biocLite warning

[Vincent Carey]

IMHO you should not be sourcing/calling biocLite in this way.  You
could use require(annotpkg, character.only=TRUE) (I think) and issue
an error if it fails, indicating to the user that they need to install
annotpkg and, if you like, suggesting to them that they use biocLite.
There are other packages that do real-time installation of, for
example, CDFs, when such are lacking, so your approach has precedents,
but I believe that embedding automatic software installation in user
applications causes more headaches than it remedies, particularly when
missing packages are large or not properly detected, or network
connectivity is lacking.

Furthermore I believe this message on visible global is not a warning
but a NOTE.  warnings must be eliminated from released packages but
NOTEs are not required to be eliminated ... for the moment.

On Fri, May 6, 2011 at 5:20 AM, swaraj basu <projectbasu at gmail.com> wrote:
> Dear All,
>             I am trying to build a package for a set of functions.  I am
> able to build the package and its working fine. When I check it with
>             R CMD check
>
>             I get a following warning :  no visible global function
> definition for ‘biocLite’
>
>              I have used biocLite to load a user defined library from
> within a function if that library is not pre-installed
>
>             if(is.element(annotpkg, installed.packages()[,1]) == "FALSE"){
>               source("http://www.bioconductor.org/biocLite.R")
>               biocLite(annotpkg)
>               library(annotpkg,character.only=TRUE)
>             }
>
>            Should I ignore this error or there is a workaround for the
> warning. My package is working fine though still I guess the warning has to
> have significance. Please help in clarifying this
>            warning.
>
>        [[alternative HTML version deleted]]
>
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
>