[Bioc-devel] RPMs for Bioconductor from Biopackages

[Martin Maechler]

>>>>> "Allen" == Allen Day <allenday at ucla.edu>
>>>>>     on Thu, 25 Jan 2007 18:24:11 -0800 writes:

    Allen> Hi,
    Allen> This is a cross post to both the Biopackages discussion list and the
    Allen> Bioconductor devel list.

    Allen> I'm an RPM developer for the Biopackages project (
    Allen> http://biopackages.net ).  We've previously built Bioconductor 1.8 for
    Allen> several different flavors and architectures of Fedora and CentOS.
    Allen> This is handled by a SGE based build farm that we've developed for the
    Allen> Biopackages project, article forthcoming.

    Allen> Now we want to package the Bioconductor 2.0 devel packages.

and you know that this implies its users use R-devel aka
 "R version x.y.z Under development (unstable) ......
		  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Of course that's nice for bioconductor *developers* who are
themselves willing to work with a bleeding-edge version of R
which can well be broken at times.

But packaging these things, means putting it into hands of useRs
who may be pretty clueless about the "unstable" part of the
underlying R version.  

Hence, (as an R-core developer) I'm a bit nervous about the
dangers of such a deployment..

Regards,
Martin Maechler, ETH Zurich


    Allen> It's a ton of work, there are ~200 of them.  I'm hoping this mail will
    Allen> stimulate collaboration between the two projects.  Perhaps we can use
    Allen> some of the code that does the daily Bioconductor testing in our build
    Allen> farm, especially if it includes code for building dependency graphs
    Allen> from the R modules, that is really the rate limiting step to building
    Allen> the RPMs because we're currently encoding dependencies manually.

    Allen> Hit me on either list or at allenday at ucla.edu if you'd like to
    Allen> participate in building RPMs for Bioconductor.  Thanks.

    Allen> -Allen