Re: [R-pkg-devel] Do I need to namespace and import functions from 'methods' package?
On 12/07/2015 4:01 PM, Dean Attali wrote: I have a package that is using the 'is' function from the 'methods' package. My package has been on CRAN for some time now and has had 3 different versions already. Yesterday when submitting, I got a response with: * checking R code for possible problems ... NOTE ggMarginal: no visible global function definition for 'is' Undefined global functions or variables: is So I fixed it by changing 'is' to 'methods::is', and then I got a response * checking dependencies in R code ... WARNING '::' or ':::' import not declared from: ‘methods’ So I suppose I need to declare 'methods' in my imported packages. My question is: is this a new requirement, to explicitly import 'methods' and namespace functions from it, or has it always been a requirement but I just wasn't caught? Or is it not even a requirement and the R CMD check was being weird? It is a new requirement. As Joshua Ulrich posted on Friday, Yes, this is expected. See: http://developer.r-project.org/blosxom.cgi/R-devel/NEWS/2015/06/29#n2015-06-29 And an earlier email on this list: https://stat.ethz.ch/pipermail/r-package-devel/2015q2/000171.html __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
Re: [R-pkg-devel] Do I need to namespace and import functions from 'methods' package?
I see there have been several messages about this recently. I apologize, I wasn't aware of the fact that I could search old messages on the board - will do that next time, thanks --- http://deanattali.com On 12 July 2015 at 13:36, Duncan Murdoch murdoch.dun...@gmail.com wrote: On 12/07/2015 4:01 PM, Dean Attali wrote: I have a package that is using the 'is' function from the 'methods' package. My package has been on CRAN for some time now and has had 3 different versions already. Yesterday when submitting, I got a response with: * checking R code for possible problems ... NOTE ggMarginal: no visible global function definition for 'is' Undefined global functions or variables: is So I fixed it by changing 'is' to 'methods::is', and then I got a response * checking dependencies in R code ... WARNING '::' or ':::' import not declared from: ‘methods’ So I suppose I need to declare 'methods' in my imported packages. My question is: is this a new requirement, to explicitly import 'methods' and namespace functions from it, or has it always been a requirement but I just wasn't caught? Or is it not even a requirement and the R CMD check was being weird? It is a new requirement. As Joshua Ulrich posted on Friday, Yes, this is expected. See: http://developer.r-project.org/blosxom.cgi/R-devel/NEWS/2015/06/29#n2015-06-29 And an earlier email on this list: https://stat.ethz.ch/pipermail/r-package-devel/2015q2/000171.html [[alternative HTML version deleted]] __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
Re: [R-pkg-devel] R-devel problem with temporary files or decompression?
I don't see any particular wrong with your code, so it seems like R.utils doesn't handle your 'file' / 'tmp' pathnames correctly. However, I cannot reproduce this (know what 'file' and 'tmp' are would help), so I cannot fix it either. Either way, instead of using low-level decompressFile(), you can achieve what you want with: rsk - R.utils::gunzip(file, temporary=TRUE, remove=FALSE) If you don't use temporary=TRUE, then 'rsk' with point to 'sample.rsk' in the current directory. /Henrik (author of R.utils) On Sun, Jul 12, 2015 at 8:45 AM, Daniel Kelley dan.kel...@dal.ca wrote: Uwe, thanks very much. How silly of me. It was because my code creates full pathnames except if the first char of a filename is “/“ or “~”, etc. I have never really used windows/msdos (went from 1970s unix to OSX) and so I was being very silly. Dan. On Jul 12, 2015, at 12:38 PM, Uwe Ligges lig...@statistik.tu-dortmund.de wrote: On 12.07.2015 17:33, Daniel Kelley wrote: A data file that I’d like to use in my “oce” package is quite large, and compresses well, so I thought of putting a compressed version into inst/extdata. I want my documentation to show how to uncompress the file, partly because these files tend to compress so well that I think users might like to store them in gzipped form. Naturally, I want to test my code using R-devel in addition to the ‘release’ version that I have on my box. Since I’ve never been able to get R-devel installed properly on OSX, I thought of submitting to win-builder. I got errors as below my signature line. What I don’t understand is the “Invalid argument”, which seems to be different from the “No such file or directory” that I would expect to get if there were no file. Any ideas on how to proceed? Dan Kelley / Oceanography / Dalhousie University / Halifax NS Canada ### ** Examples library(oce) ## A CTD-style rsk if (requireNamespace(R.utils)) { + file - system.file(extdata, sample.rsk.gz, package=oce) + tmp - tempfile(fileext=.rsk) + R.utils::decompressFile(file, destname=tmp, ext=gz, FUN=gzfile, remove=FALSE) + rsk - read.rsk(tmp) + ctd - as.ctd(rsk) + plot(ctdTrim(ctd)) + } Loading required namespace: R.utils Warning in file(file, r) : cannot open file 'd:/RCompile/CRANguest/R-devel/oce.Rcheck/examples_i386/D:\temp\RtmpGUYumI\file29ac3bed3fed.rsk': Apparently you have put together the location of the current path and the one of the tempfile which does not make sense. So very likely a bug in your code. Surprise you do not see it on your end. Best, Uwe Ligges Invalid argument Error in file(file, r) : cannot open the connection Calls: read.rsk - file Execution halted ** running examples for arch 'x64' ... ERROR __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
[R-pkg-devel] Do I need to namespace and import functions from 'methods' package?
I have a package that is using the 'is' function from the 'methods' package. My package has been on CRAN for some time now and has had 3 different versions already. Yesterday when submitting, I got a response with: * checking R code for possible problems ... NOTE ggMarginal: no visible global function definition for 'is' Undefined global functions or variables: is So I fixed it by changing 'is' to 'methods::is', and then I got a response * checking dependencies in R code ... WARNING '::' or ':::' import not declared from: ‘methods’ So I suppose I need to declare 'methods' in my imported packages. My question is: is this a new requirement, to explicitly import 'methods' and namespace functions from it, or has it always been a requirement but I just wasn't caught? Or is it not even a requirement and the R CMD check was being weird? --- http://deanattali.com [[alternative HTML version deleted]] __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
Re: [R-pkg-devel] R-devel problem with temporary files or decompression?
Uwe, thanks very much. How silly of me. It was because my code creates full pathnames except if the first char of a filename is “/“ or “~”, etc. I have never really used windows/msdos (went from 1970s unix to OSX) and so I was being very silly. Dan. On Jul 12, 2015, at 12:38 PM, Uwe Ligges lig...@statistik.tu-dortmund.de wrote: On 12.07.2015 17:33, Daniel Kelley wrote: A data file that I’d like to use in my “oce” package is quite large, and compresses well, so I thought of putting a compressed version into inst/extdata. I want my documentation to show how to uncompress the file, partly because these files tend to compress so well that I think users might like to store them in gzipped form. Naturally, I want to test my code using R-devel in addition to the ‘release’ version that I have on my box. Since I’ve never been able to get R-devel installed properly on OSX, I thought of submitting to win-builder. I got errors as below my signature line. What I don’t understand is the “Invalid argument”, which seems to be different from the “No such file or directory” that I would expect to get if there were no file. Any ideas on how to proceed? Dan Kelley / Oceanography / Dalhousie University / Halifax NS Canada ### ** Examples library(oce) ## A CTD-style rsk if (requireNamespace(R.utils)) { + file - system.file(extdata, sample.rsk.gz, package=oce) + tmp - tempfile(fileext=.rsk) + R.utils::decompressFile(file, destname=tmp, ext=gz, FUN=gzfile, remove=FALSE) + rsk - read.rsk(tmp) + ctd - as.ctd(rsk) + plot(ctdTrim(ctd)) + } Loading required namespace: R.utils Warning in file(file, r) : cannot open file 'd:/RCompile/CRANguest/R-devel/oce.Rcheck/examples_i386/D:\temp\RtmpGUYumI\file29ac3bed3fed.rsk': Apparently you have put together the location of the current path and the one of the tempfile which does not make sense. So very likely a bug in your code. Surprise you do not see it on your end. Best, Uwe Ligges Invalid argument Error in file(file, r) : cannot open the connection Calls: read.rsk - file Execution halted ** running examples for arch 'x64' ... ERROR __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel