Yeah, this is not very clear, although repeated posts on R-devel suggests that R-core believes it ought to be common knowledge to always run R CMD INSTALL/check on a tarball produced by R CMD build.
I see Martin is reporting this on R-devel. Best, Kasper On Mon, Oct 21, 2013 at 2:37 AM, Gordon K Smyth <[email protected]> wrote: > Thanks, setting pkg to be the tar.gz file instead of the directory does > the job. I had not appreciated the difference. > > The code on checkResults is very clear now. I hadn't appreciated that the > BUILD and the BUILDBIN steps were cumulative. > > Thanks > Gordon > > > > On Sun, 20 Oct 2013, Martin Morgan wrote: > > On 10/20/2013 10:36 PM, Gordon K Smyth wrote: >> >>> Hi Martin, >>> >>> Thanks for the heads-up. I think I've made the changes necessary, but >>> am having trouble confirming whether the changes are correct on a Windows >>> machine. >>> >>> Could you tell me hows Bioconductor builds the Windows binaries of >>> Bioconductor packages? >>> >>> To see how things work, I am playing around with Biobase on svn. I can >>> build Biobase from the svn source including the pdfs of the vignettes using >>> >>> R CMD build Biobase >>> >>> on my Windows machine. This gives me a tar.gz file of Biobase with an >>> inst/doc directory containing Rnws and pdfs of all the vignettes. This >>> shows that my LaTeX etc is working fine. >>> >>> However when I run >>> >>> R CMD INSTALL --build Biobase >>> >>> which is the recommended way to install a package under Windows, I get a >>> zip file containing no sign of any vignette. In fact there is no doc or >>> inst or vignette directory in the zip file. >>> >>> What else do I need to do to install the package from source with the >>> vignettes build under Windows? >>> >>> No doubt the answer is somewhere in one of the R manuals (writing R >>> extensions or admin and installation perhaps) but I cannot find it. >>> >> >> The build machines run >> >> R CMD INSTALL --build Biobase_2.23.1.tar.gz >> >> i.e., on the file created by R CMD build Biobase. This is not >> inconsistent with 1.3.3 of Writing R Extensions: >> >> "R CMD INSTALL --build pkg where pkg is either the name of a source >> tarball (in the usual .tar.gz format) or the location of the directory of >> the package source to be built. " >> >> although there really seems to be only one way to build complete >> binaries. The Bioconductor build system reports what it's doing in the >> 'Summary' section 'Command' line at >> >> >> http://bioconductor.org/**checkResults/devel/bioc-** >> LATEST/Biobase/moscato2-**buildbin.html<http://bioconductor.org/checkResults/devel/bioc-LATEST/Biobase/moscato2-buildbin.html> >> >> Martin >> >> >>> Thanks >>> Gordon >>> >>> PS. I am running Rtools3.1 with: >>> >>> sessionInfo() >>>> >>> R Under development (unstable) (2013-10-18 r64076) >>> Platform: i386-w64-mingw32/i386 (32-bit) >>> >>> locale: >>> [1] LC_COLLATE=English_Australia.**1252 >>> [2] LC_CTYPE=English_Australia.**1252 >>> [3] LC_MONETARY=English_Australia.**1252 >>> [4] LC_NUMERIC=C >>> [5] LC_TIME=English_Australia.1252 >>> >>> attached base packages: >>> [1] parallel stats graphics grDevices utils >>> [6] datasets methods base >>> >>> other attached packages: >>> [1] Biobase_2.23.1 BiocGenerics_0.9.0 >>> >>> >>> >>> >>> Date: Fri, 18 Oct 2013 14:58:52 -0700 >>>> From: Martin Morgan <[email protected]> >>>> To: "[email protected]" <[email protected]> >>>> Subject: [Bioc-devel] IMPORTANT inst/doc/*Rnw to vignettes/ >>>> >>>> Bioconductor developers, >>>> >>>> Thank you for your contributions to our last release, with a total of >>>> 749 >>>> packages! This is a great resource for the bioninformatics community. >>>> >>>> No good deed goes unpunished. The 'devel' version of R no longer builds >>>> vignettes in the inst/doc directory. Vignette sources (.Rnw or perhaps .Rmd >>>> files) must be in the vignettes/ directory. There's a lot of work to be >>>> done to make this happen, and we'd really like your help with this. Here's >>>> one recipe for the neccessary changes, illustrtated with Biobase >>>> >>>> 1. Find the url of your package in the devel version of the >>>> Bioconductor svn repository, e.g., for Biobase >>>> >>>> https://hedgehog.fhcrc.org/**bioconductor/trunk/madman/**Rpacks/Biobase<https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/Biobase> >>>> >>>> 2. Make a clean check-out of the source into a directory where >>>> `Biobase' does not already exist >>>> >>>> svn co https://hedgehog.fhcrc.org/**bioconductor/trunk/madman/** >>>> Rpacks/Biobase<https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/Biobase> >>>> >>>> 3. Re-name the inst/doc directory. Using svn rename preserves the >>>> commit history >>>> >>>> svn rename Biobase/inst/doc Biobase/vignettes >>>> >>>> 4. Increase the least significant part of the package version number in >>>> Biobase/DESCRIPTION. I changed Biobase from 2.23.0 to 2.23.1. >>>> >>>> 5. Build the package, using the latest devel version of R >>>> >>>> R CMD build Biobase >>>> >>>> 6. Check that the built version contains .Rnw, .R, and .pdf files in the >>>> inst/doc directory, e.g., >>>> >>>> tar tzf Biobase_2.23.1.tar.gz |grep inst/doc >>>> >>>> 7. Make sure the built package passes R CMD check >>>> >>>> R CMD check Biobase_2.23.1.tar.gz >>>> >>>> 8. Review and check in your changes >>>> >>>> svn st Biobase >>>> svn diff Biobase/DESCRIPTION >>>> svn ci >>>> >>>> This will work for many cases, but if you have problems please do not >>>> hesitate to bring up issues on the mailing list. >>>> >>>> If things go smoothly, you might wish to follow the instructions in the >>>> BiocStyle vignette >>>> >>>> >>>> http://bioconductor.org/**packages/devel/bioc/vignettes/** >>>> BiocStyle/inst/doc/LatexStyle.**pdf<http://bioconductor.org/packages/devel/bioc/vignettes/BiocStyle/inst/doc/LatexStyle.pdf> >>>> >>>> >>>> for creating vignettes with a consistent look-and-feel. The text is >>>> wider, for instance, so code generally formats better, and packages are >>>> easily linked to their 'landing page' on the Bioconductor web site >>>> >>>> Please don't hesitate to ask if there are any questions, and thanks in >>>> advance for your continued effort to support the project. >>>> >>>> Martin >>>> -- >>>> Computational Biology / Fred Hutchinson Cancer Research Center >>>> 1100 Fairview Ave. N. >>>> PO Box 19024 Seattle, WA 98109 >>>> >>>> Location: Arnold Building M1 B861 >>>> Phone: (206) 667-2793 >>>> >>>> >>> >> >> -- >> Computational Biology / Fred Hutchinson Cancer Research Center >> 1100 Fairview Ave. N. >> PO Box 19024 Seattle, WA 98109 >> >> Location: Arnold Building M1 B861 >> Phone: (206) 667-2793 >> >> > ______________________________**______________________________**__________ > The information in this email is confidential and inte...{{dropped:11}} _______________________________________________ [email protected] mailing list https://stat.ethz.ch/mailman/listinfo/bioc-devel
