Hi, we've got a local mirror of Bioconductor (from version 2.1 to 2.3) and I'd like to know to do such that any request to the bioconductor repository (by any package) is actually send to our local mirror. We setup a system-wide 'repos' options so that when a user do install.packages, the local mirror is included in the searched repositories. BUT: - for example the GCRMA package (in 2.1 at least) seems to get its reposiroty URL from getOption('BioC'), which still point to bioconductor.org - a call to Biobase::repoList() also returns bioconductor.org URLs Is there a recommended way to change that so that the user doesn't have to car about the repositories, i.e. something that makes Biobase::repoList() returns our local mirror directories. Something to configure at the installation maybe, or override the function in the Rprofile.site file? Obviously, if some packages have hard coded URLs then it won't work, but I guess this is part of the things that are checked before a package is actually included in the repository. Thanks

Hi Renaud, Thanks for mirroring Bioconductor! I would be glad to add you to this page http://bioconductor.org/download/mirrors/ if you wanted (just send me the URL + the name of your organization). I've modified the Biobase::biocReposList() function in the release and devel versions of Biobase (2.2.1 and 2.3.2) so now it gets the top URL from getOption("BioC"). Also loading Biobase should not overwrite the user defined "BioC" option anymore. This change will propagate to the public repos in about 24h. I agree that it would be nice if all the packages that need to access the Bioconductor repos didn't use a hard-coded URL for this but we cannot guarantee this (biocReposList() itself was doing this until a few minutes ago :-/). That would require us to review all the current packages. It will be easier if people report the problem here on the list or directly with the package maintainer every time this problem is found (there might not be that many packages that try to access the BioC repos). We'll add this requirement for future contributed packages though. Other people are mirroring Bioconductor and surely have encountered this problem too. I guess you've already done this but, just in case, biocLite() first needs to be installed locally and modified so it points to the local mirror. The biocLite() function is defined in a set of stand-alone scripts (they are not part of any package) that you can get from svn here: https://hedgehog.fhcrc.org/bioconductor/trunk/madman/installScripts You need to put the installScripts/ folder somewhere on your website, eventually create a symlink at the top level of your website to point to the biocLite.R script, and modify biocLite.R + the files under installScripts/2.8 and installScripts/2.9 to make them point to your local mirror. Please let us know if you need further help. Thanks! H. Renaud Gaujoux wrote: > Hi, > > we've got a local mirror of Bioconductor (from version 2.1 to 2.3) and > I'd like to know to do such that any request to the bioconductor > repository (by any package) is actually send to our local mirror. > We setup a system-wide 'repos' options so that when a user do > install.packages, the local mirror is included in the searched > repositories. > BUT: > - for example the GCRMA package (in 2.1 at least) seems to get its > reposiroty URL from getOption('BioC'), which still point to > bioconductor.org > - a call to Biobase::repoList() also returns bioconductor.org URLs > > Is there a recommended way to change that so that the user doesn't have > to car about the repositories, i.e. something that makes > Biobase::repoList() returns our local mirror directories. Something to > configure at the installation maybe, or override the function in the > Rprofile.site file? > > Obviously, if some packages have hard coded URLs then it won't work, but > I guess this is part of the things that are checked before a package is > actually included in the repository. > > Thanks > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor