Hello, I don't think CRAN actually is a problem. They have archives available (+ recent investigation lead me to the conclusion that proxy I'm forced to usr is -- how do I put it lightly? -- COMPLETE AND UTTER BULLCRAP).
What is more of my concern is BioConductor. They don't have archives and it seems they don't consider adding one. Consider the packages broken or not I need to have a stable archive available so the (my) company needs to invest some money for the storage or live with the way its provided. I'm thinking about a patch in the terms of "--print-uris" to circumvent that (might also be nice for PyPi, CPAN, ...). Once the sources are at the local site I (ay least) have no one to blame but myself... /Martin On Wed, Apr 1, 2015, 14:42 Jens Timmerman <jens.timmer...@ugent.be> wrote: > Hi all, > > I just came about revoultionary R, which takes a CRAN snapshot for every > release they do, > this could provide us with a lot more stability in the packages if we > follow these repository snapshots. > > > http://mran.revolutionanalytics.com/documents/rro/reproducibility/#reproducibility > > > Regards, > Jens Timmerman > > On 31/03/15 18:26, Martin wrote: > > How about getting rid of of versions (at least for Bioconductor) and just > let R itself deal with it. I find that "BiocInstaller::biocLite(....)" does > work quite nicely. Be it an upstream mirror or a site local mirror, with a > site local mirror the set of packages is under the local of the the site > and there's no need to remove them during a mirror sync so one would at > least have some kind of archive... > > Of course that may then introduce other things like: "Is there a smart > way to set an Rprofile.site from within an easyconfig?" (Yes I'm being > greedy, EB in general works well enough for me to worry about these Luxury > problems) > > /Martin > > > On Tue, Mar 31, 2015 at 6:15 PM Martin <mar...@marcher.name> wrote: > >> Hi, >> >> Yes I know about -k it just seems so cumbersome to do that :( >> >> The combination of a broken corporate proxy and files that go missing >> really makes this a mess (for me, EB itself is great, I just have to suffer >> the slow Enterprise Death). >> >> I'll look into the script, thanks. >> >> Martin >> >> On Tue, Mar 31, 2015 at 5:58 PM Pablo Escobar Lopez < >> pablo.escobarlo...@unibas.ch> wrote: >> >>> Hi Martin, >>> >>> R extensions change quite often and the url also changes so the normal >>> approach is "manual fix". In case it's useful for you I wrote this script >>> some days ago to deal with this, it's not perfect and still requires some >>> manual work but it works for ~90% of the R libraries so it saves you some >>> time >>> >>> https://github.com/pescobar/random-scripts/blob/master/easybuild-update-R-libs.py >>> >>> To install extra R libraries you can just add the new library to the >>> exts_list and then rerun the easyconfig with "eb R-3.1.2.eb -k -f" This >>> way only libraries not already installed will be compiled. This is commonly >>> used to add extra libraries for R/Perl/Python. This is the easybuild help: >>> >>> -k, --skip Skip existing software (useful for installing >>> additional packages) (def False) >>> >>> >>> regards, >>> pablo. >>> >>> 2015-03-31 17:41 GMT+02:00 Martin <mar...@marcher.name>: >>> >>>> Hi, >>>> >>>> I'm trying to install R-3.1.2-goolf-1.5.14.eb. >>>> >>>> I'm running a simple: >>>> >>>> eb R-3.1.2-goolf-1.5.14.eb --robot -s fetch >>>> >>>> but it fails on most of the packages in exts_list. How are people >>>> dealing with this? >>>> >>>> Are you just going thru the list and manually fix the versions, I >>>> keep thinking there has to be a better way. >>>> >>>> Also how do you (later) install an additional package? Extend the >>>> list in exts_list and "reinstall" or provide another module? Are there >>>> established best practices for this? >>>> >>>> thanks, >>>> Martin >>>> >>>> >>>> >>> >>> >>> -- >>> Pablo Escobar López >>> HPC systems engineer >>> Biozentrum, University of Basel >>> Swiss Institute of Bioinformatics SIB >>> Email: pablo.escobarlo...@unibas.ch >>> Phone: +41 61 267 21 80 >>> http://www.biozentrum.unibas.ch >>> >>
