the issue posted https://github.com/hadley/devtools/issues/864
On Fri, Jun 26, 2015 at 2:36 PM, Vladislav Petyuk <pet...@gmail.com> wrote: > It looks like devtools::package_deps is forcing type to be "binary". > if (identical(type, "both")) { > type <- "binary" > } > > > > > devtools::package_deps > function (pkg, dependencies = NA, repos = getOption("repos"), > type = getOption("pkgType")) > { > if (identical(type, "both")) { > type <- "binary" > } > if (length(repos) == 0) > repos <- character() > repos[repos == "@CRAN@"] <- "http://cran.rstudio.com" > cran <- available_packages(repos, type) > if (missing(pkg)) { > pkg <- as.package(".")$package > } > deps <- sort(find_deps(pkg, cran, top_dep = dependencies)) > inst <- installed.packages() > base <- unname(inst[inst[, "Priority"] %in% c("base", "recommended"), > "Package"]) > deps <- setdiff(deps, base) > inst_ver <- unname(inst[, "Version"][match(deps, rownames(inst))]) > cran_ver <- unname(cran[, "Version"][match(deps, rownames(cran))]) > diff <- compare_versions(inst_ver, cran_ver) > structure(data.frame(package = deps, installed = inst_ver, > available = cran_ver, diff = diff, stringsAsFactors = FALSE), > class = c("package_deps", "data.frame"), repos = repos, > type = type) > } > <environment: namespace:devtools> > > > On Fri, Jun 26, 2015 at 2:12 PM, Vladislav Petyuk <pet...@gmail.com> > wrote: > >> Dan, Herve, thanks for taking a look. This gives some leads. >> >> On Thu, Jun 25, 2015 at 7:26 PM, Dan Tenenbaum <dtene...@fredhutch.org> >> wrote: >> >>> >>> >>> ----- Original Message ----- >>> > From: "Hervé Pagès" <hpa...@fredhutch.org> >>> > To: "Vladislav Petyuk" <pet...@gmail.com>, "Dan Tenenbaum" < >>> dtene...@fredhutch.org> >>> > Cc: Bioc-devel@r-project.org >>> > Sent: Thursday, June 25, 2015 6:56:56 PM >>> > Subject: Re: [Bioc-devel] Challenges with package installation on >>> fresh R >>> > >>> > Hi Vladislav, >>> > >>> > On 06/25/2015 01:13 PM, Vladislav Petyuk wrote: >>> > > Yes, the issue is the same if I use >>> > > devtools::install_github("vladpetyuk/toypack", >>> > > build_vignettes=TRUE, >>> > > dependencies=TRUE) >>> > > >>> > > I posted on Bioc-devel list because all the problematic packages >>> > > are >>> > > suspiciously similar - all related to Bioconductor and deal with >>> > > some sort >>> > > of databases (GO/KEGG/Reactome). >>> > >>> > What all these packages have in common is that we don't provide >>> > binaries >>> > for them anymore. That's because install.packages() (and thus >>> > biocLite(), which uses install.packages() internally) knows how to >>> > deal with this (via type="both", which is now the default on Windows >>> > and Mac). However, some core utilities like >>> > utils::download.packages() >>> > have not been properly adapted to handle the situation where a binary >>> > is missing. For example with R-3.2.1 on Windows: >>> > >>> > > library(BiocInstaller) >>> > > download.packages("GO.db", destdir=".", >>> > > repos=biocinstallRepos()) >>> > trying URL >>> > ' >>> http://bioconductor.org/packages/3.2/data/annotation/src/contrib/GO.db_3.1.2 >>> ' >>> > Error in download.file(url, destfile, method, mode = "wb", ...) : >>> > cannot open URL >>> > ' >>> http://bioconductor.org/packages/3.2/data/annotation/src/contrib/GO.db_3.1.2 >>> ' >>> > In addition: Warning message: >>> > In download.file(url, destfile, method, mode = "wb", ...) : >>> > cannot open: HTTP status was '404 Not Found' >>> > Warning in download.packages("GO.db", destdir = ".", repos = >>> > biocinstallRepos()): >>> > download of package 'GO.db' failed >>> > [,1] [,2] >>> > >>> > This could be what breaks install_github() (which doesn't seem to >>> > use install.packages() internally but to instead handle the >>> > downloading >>> > of the packages itself). >>> >>> install_github() calls install() whose documentation says that its ... >>> arguments are passed to install.packages() in order to install >>> dependencies. But I agree that something else seems to be going on here and >>> that devtools decides the packages can't be installed without running >>> install.packages(). >>> >>> Dan >>> >>> >>> > >>> > Cheers, >>> > H. >>> > >>> > > >>> > > On Thu, Jun 25, 2015 at 1:08 PM, Dan Tenenbaum >>> > > <dtene...@fredhutch.org> >>> > > wrote: >>> > > >>> > >> >>> > >> >>> > >> ----- Original Message ----- >>> > >>> From: "Vladislav Petyuk" <pet...@gmail.com> >>> > >>> To: "Dan Tenenbaum" <dtene...@fredhutch.org> >>> > >>> Cc: Bioc-devel@r-project.org >>> > >>> Sent: Thursday, June 25, 2015 12:55:56 PM >>> > >>> Subject: Re: [Bioc-devel] Challenges with package installation on >>> > >>> fresh R >>> > >>> >>> > >>> >>> > >>> Just added the full output to the README.md file >>> > >>> https://github.com/vladpetyuk/toypack/blob/master/README.md >>> > >>> >>> > >>> >>> > >> >>> > >> Do you get the same results if you do: >>> > >> >>> > >> devtools::install_github("vladpetyuk/toypack", >>> > >> build_vignettes=TRUE, >>> > >> dependencies=TRUE) >>> > >> >>> > >> ? >>> > >> >>> > >> When you pass biocLite() a username/repos argument as you've done, >>> > >> it just >>> > >> delegates all the work to devtools::install_github(). >>> > >> >>> > >> If you get the same issue with install_github(), then the issue >>> > >> has >>> > >> nothing to do with Bioconductor and you need to discuss it with >>> > >> the >>> > >> maintainer of devtools. >>> > >> If not, then maybe there is an issue with the way we delegate to >>> > >> that >>> > >> command. >>> > >> >>> > >> Dan >>> > >> >>> > >> >>> > >>> >>> > >>> >>> > >>> >>> > >>> On Thu, Jun 25, 2015 at 12:28 PM, Dan Tenenbaum < >>> > >>> dtene...@fredhutch.org > wrote: >>> > >>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> ----- Original Message ----- >>> > >>>> From: "Vladislav Petyuk" < pet...@gmail.com > >>> > >>>> To: Bioc-devel@r-project.org >>> > >>>> Sent: Thursday, June 25, 2015 12:16:50 PM >>> > >>>> Subject: [Bioc-devel] Challenges with package installation on >>> > >>>> fresh >>> > >>>> R >>> > >>>> >>> > >>>> Hi all, >>> > >>>> I ran into a problem with dependencies during package >>> > >>>> installation. >>> > >>>> Here is an example that intends to reproduce the problem >>> > >>>> >>> > >>>> https://github.com/vladpetyuk/toypack >>> > >>>> >>> > >>>> Majority of the packages (in Imports/Depends/Suggests and their >>> > >>>> Imports/Depends/Suggests) install just fine. >>> > >>>> However there is a handful of packages ("DO.db", "GO.db", >>> > >>>> "org.Ce.eg.db", >>> > >>>> "org.Hs.eg.db" and "reactome.db") that are being skipped for not >>> > >>>> an >>> > >>>> obvious >>> > >>>> to me reason. >>> > >>>> In this toy example there is no use of those skipped packages. >>> > >>>> So >>> > >>>> it >>> > >>>> won't >>> > >>>> results in any kind of error. But in the real package their lack >>> > >>>> results >>> > >>>> in errors. >>> > >>>> >>> > >>> >>> > >>> You need to provide the complete output of R when you try and >>> > >>> install >>> > >>> the package. >>> > >>> The error is in that output somewhere. >>> > >>> >>> > >>> Dan >>> > >>> >>> > >>> >>> > >>>> Thank you, >>> > >>>> Vlad >>> > >>>> >>> > >>>> [[alternative HTML version deleted]] >>> > >>>> >>> > >>>> _______________________________________________ >>> > >>>> Bioc-devel@r-project.org mailing list >>> > >>>> https://stat.ethz.ch/mailman/listinfo/bioc-devel >>> > >>>> >>> > >>> >>> > >>> >>> > >> >>> > > >>> > > [[alternative HTML version deleted]] >>> > > >>> > > _______________________________________________ >>> > > Bioc-devel@r-project.org mailing list >>> > > https://stat.ethz.ch/mailman/listinfo/bioc-devel >>> > > >>> > >>> > -- >>> > Hervé Pagès >>> > >>> > Program in Computational Biology >>> > Division of Public Health Sciences >>> > Fred Hutchinson Cancer Research Center >>> > 1100 Fairview Ave. N, M1-B514 >>> > P.O. Box 19024 >>> > Seattle, WA 98109-1024 >>> > >>> > E-mail: hpa...@fredhutch.org >>> > Phone: (206) 667-5791 >>> > Fax: (206) 667-1319 >>> > >>> >> >> > [[alternative HTML version deleted]] _______________________________________________ Bioc-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/bioc-devel