[BioC] bioconductor and ESS

[Barry Rowlingson]

A.J. Rossini wrote:
> I really think Wolfgang is correct -- that particular command
> shouldn't be doing anything to change results within or without of
> Emacs.

>>are you sure that you are actually using the same versions of R when you
>>call Rgui versus ESS? Perhaps your emacs accesses an old version of R
>>that is still lying around somewhere on your computer.

  Its the same binary, Patrick has it in D:\program 
files\R\bin\Rterm.exe - if we run that from the DOS command prompt it 
works, when launched from ESS (and I checked the ESS setup, and it does 
run that .exe file) it has the problem.

  The problem can be seen by:

library(Biobase)
getClassDef(class(new("phenoData")))@validity

  which gives this in ESS:

function (object)
{
    dm <- dim(object at pData)
    if (dm[2] != length(object at phenoLabels))
        return(FALSE)
    return(TRUE)
}
<environment: 00A50680>

  - and then object at phenoLabels fails, since these objects dont have 
those slots. The validity method from Rterm or Rgui checks the correct 
slot.

  I suspect there must be an old version of some bioconductor packages 
around somewhere, but I dont know why ESS is using them. .libPaths() is 
identical in all environments.

  Cant give any more specifics now, since Patrick has gone off to watch 
mountaineer Chris Bonington become our new Chancellor...

Baz