[Rd] ref #2655 (PR#3566)

[Prof Brian Ripley]

There is nothing here to reproduce!  Please do not use R-bugs for help 
requests (and please do use a meaningful subject line).  Also, R-bugs is 
not the place to report bugs (if you find one) in contributed packages, 
and stepAIC is in a contributed package (MASS).  You have not even told us 
the version of that package which you have used.  [All this is in the FAQ 
and in ?bug.report, so please read them. Especially since you did the 
same things with PR#2655.]

This is unlikely to be a bug, more likely a limitation of R's scope rules.
If you send a reproducible example to R-help you are likely to get help 
about how to attack your problem in R.

On Sat, 26 Jul 2003 slteng at stat.berkeley.edu wrote:

> Full_Name: Melinda Teng
> Version: 1.7.1
> OS: Windows ME
> Submission from: (NULL) (67.118.2.50)
> 
> 
> Hi,
> 
> I understand that R 1.7.0 version (onwards) will solve the issue in #2655.
> However, I still find the same error message. That is, a data matrix defined
> outside the function containing the stepAIC() (alongwith glm()) is not passed
> into the function. This data matrix is defined as one of the function
> parameters.
> 
> Error in model.frame.default(formula = Y ~ X4, data = A, drop.unused.levels =
> TRUE) : 
>         Object "A" not found

In that usage A should be a data frame, not a matrix, BTW.

-- 
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