[Rd] new warnings related to the extractor "$" with R 2.5.0alpha

[Prof Brian Ripley]

On Sun, 1 Apr 2007, Renaud Lancelot wrote:

> Dear all,
>
> I just installed R 2.5.0alpha and noticed new warnings related to the
> extractor "$" when using contributed packages. For instance:
>
>> library(RODBC)
> Warning message:
> $ operator is not valid for atomic vectors, returning NULL
>
>> library(aod)
> Package aod, version 1.1-18
>> data(orob2)
>> m1 <- betabin(cbind(y, n-y) ~ 1, random = ~ 1, data = orob2)
>> head(residuals(m1))
>         1          2          3          4          5          6
> -1.4629774 -0.7897877 -1.3754283  0.1029910 -0.3556985  1.4030026
> Warning message:
> $ operator not defined for this S4 class, returning NULL
>
> In these two cases, I deleted the former installations and re-built
> the packages from sources.
>
> How to overcome this problem ?

Wait for package updates.  For RODBC, it is harmless but the warning will 
not appear in 1.1-9 (which will be released before R 2.5.0 is).

(The first message is not really right: $ for lists was valid and 
documented to return NULL in released versions of R, as it was in the Blue 
Book.  It should say that it is 'deprecated' in R 2.5.0, since this 
amounts to a change of language definition and that should be giving a 
warning of deprecation.)

Later versions of R 2.5.0 alpha give more informative messages.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595