[R] use of NULL environment is deprecated?

[Duncan Murdoch]

On 7/11/2006 6:59 PM, Patrick Connolly wrote:
> On Tue, 11-Jul-2006 at 06:41PM -0400, Duncan Murdoch wrote:
> 
> [....]
> 
> |> 
> |> options(warn=2) will convert warnings into errors, so traceback will work.
> |> 
> |> A common situation where I've seen that error is with binary saves from 
> |> earlier versions of R being loaded into current versions.  For example, 
> |> if you installed a package before, but didn't re-install it with 2.3.1, 
> |> or if you are reloading a workspace saved in an earlier version.
> 
> Thank you for the explanation.  In this case, it's the latter reason:
> the workspace comes from a different computer as well as a different
> version of R.
> 
> Will saving the objects anew overcome the problem?

Yes, it should, though there's a slight risk that someone might have 
meant NULL to mean an empty environment instead of the base environment. 
  This would be a bug with or without the change.

Duncan Murdoch