[Bioc-devel] R version in Bioc packages (was Re: [BioC] Package \'jmosaics\' not found)

[Dan Tenenbaum]

On Tue, May 21, 2013 at 1:01 PM, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
> Hi Steffen,
>
>
>
> On Tue, May 21, 2013 at 12:40 PM, Steffen Neumann <sneumann at ipb-halle.de> wrote:
>> Hi,
>>
>> On Tue, 2013-05-21 at 11:47 -0400, James W. MacDonald wrote:
>>> Not that I can recall. But I do agree that it is at minimum redundant
>>> and probably just wrong to have R-version requirements for BioC packages.
>>
>> I'd like to disagree. BioC *promises* that stuff works together
>> at a certain point in time, simply because that stuff was green
>> at the build farm at that time.
>>
>
> No, it's more than that. We *intend* that all packages in a given BioC
> version will work with all other packages in that version. Conversely,
> we pretty much guarantee problems if you start mixing and matching.
>
>
>
>> But what should keep me as a developer from promising
>> that my stuff also works on older releases ?
>> (For the fun of it, I once bisected R versions to find
>>  a minimal R version for xcms ...).
>>
>
> For one thing, dependeing on when your package was introduced, there
> is an older version of it already available that will work with the
> older R version. Of course it may not have the coolest new features
> but it is there.
> For example, xcms was introduced in BioC 1.6, a long time ago. And so
> anyone running R 2.1 or earlier can simply

er, 2.1 or *later*



> source("http://bioconductor.org/biocLite.R")
> biocLite("xcms")
> and the appropriate version of xcms (and all its dependencies) for
> that version of BioC will be installed.
>
>
>
>> So why shouldn't I declare that xcms still runs on R-2.14.0 ?
>> I can R CMD check it there, and if someone knows what he does,
>> he can bypass biocLite() and install the latest from SVN manually.
>
>
> Sure....one can do a lot of things but that doesn't mean they are the
> right things to do.
>
> Dan
>
>
>>
>> If I *know* xcms won't work reliably with R-2.13 and older,
>> then omitting "Depends: R (>= 2.14.0)" means that someone
>> with an old cluster (which oftentimes receives fewer upgrades
>> than your average laptop) with venerable R-2.10.0 fill fail
>> *at some time* rather than on install time.
>>
>> Just my 2c.
>>
>> Yours,
>> Steffen
>>
>> --
>> IPB Halle                    AG Massenspektrometrie & Bioinformatik
>> Dr. Steffen Neumann          http://www.IPB-Halle.DE
>> Weinberg 3                   http://msbi.bic-gh.de
>> 06120 Halle                  Tel. +49 (0) 345 5582 - 1470
>>                                   +49 (0) 345 5582 - 0
>> sneumann(at)IPB-Halle.DE     Fax. +49 (0) 345 5582 - 1409
>>
>> _______________________________________________
>> Bioc-devel at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/bioc-devel