[BioC] getBioC(relLevel="devel") fails under r-devel

[Douglas Bates]

Jeff Gentry <jgentry at jimmy.harvard.edu> writes:

> > Using a current version of r-devel on a Debian GNU/Linux system I
> > encounter an error in getBioC(relLevel="devel").
> 
> Is this happening to you on R-release?  Is it possible that something has
> changed in the time between R-release and now in the R-devel branch that
> would be causing this?
> 
> I'm downloading R-devel now, will see what is going on.  I can't replicate
> this using R-1.8 (which for about a week longer is what the BioC-devel is
> going to be tracking instead of R-devel)

The problem existed only in my modified version of R-devel.  I had
made a modification so that assignment of an attribute did not copy
the value if NAMED(val) == 0.  Up until now we thought that this
modification would be benign and would give a performance boost to
packages that use S4 classes and methods, which means most of
Bioconductor.

However, it seems this change is not benign.  In a sense this is good
news because it will allow us to isolate where the problem lies.

I have not committed this change to r-devel so others need not worry
about the use of a current r-devel breaking Bioconductor.
-- 
Douglas Bates                            bates at stat.wisc.edu
Statistics Department                    608/262-2598
University of Wisconsin - Madison        http://www.stat.wisc.edu/~bates/