>>>>> "Cooper" == Cooper Bethea <cooper.bethea at
duke.edu>
>>>>> on 13 Jun 2003 14:28:02 -0400 writes:
Cooper> hi all- apologies if this question has come up
Cooper> before; i took a swing through a few months of
Cooper> archives and didn't turn anything up.
Cooper> i'm a linux systems administrator, and we're running
Cooper> R across a beowulf cluster. i've been asked to
Cooper> install the bioconductor package, for which i can't
Cooper> find RPMs. the recommended way to do this on the
Cooper> bioconductor home page is via R CMD INSTALL, which
Cooper> seems to be the standard way to install R packages.
Cooper> however, i want to make an RPM out of bioconductor,
Cooper> since i have to deploy it to 48 machines and i want
Cooper> rpm to track the versioning and dependencies. also,
Cooper> the machines are very homogenous and there's no
Cooper> reason compilation should occur on each.
I'm not really answering your primary question,
BUT
- don't you share some file system(s) between all clients?
If yes, you can "R CMD INSTALL -l <librarydirectory>
package"
(and make sure your users have <librarydirectory> in their
R_LIBS or use .libPaths() in the site-wide Rprofile.
See help(Startup), help(library) {which has .libPaths()}.
- if you don't share any file systems {quite improbable} AND
your machines are very homogenous, you can still
INSTALL on one machine, and use "rsync" or some other
synchronization tool to have the installation distributed
to the clients. You will do this with other stuff
(/etc/.... configurations), anyway, won't you?
Hence, no really deep need for RPMs there, I think.
Cooper> so could someone who's built R extension RPM
Cooper> packages give me some pointers? a SRPM for a package
Cooper> would be invaluable.