R help - Sep 1999 - data frame component replacement: feature or bug?

Hi, all.

The following does not behave as I think it should, and as it seems to me
it has in the past (although I can't check this easily).  I know it
happens in both R-0.64.2 and R-0.65.0 on an old Power Computing running
Linux-PPC 1999, and in R-0.64.2 on an SGI running Irix 6.5.

Try the following:

t1 <- data.frame(matrix(rnorm(16), nc=4))
> t1
          X1 X2 X3 X4 
1 -0.7206945   0.4511310  -1.0638681 -0.9128938 
2 -1.6340557  -0.8063090   0.7753397  0.3607708 
3  0.5999834   0.5256372   0.4393342  1.3218485 
4  2.8542598  -2.8740136   2.0165816 -0.3026954
> dim(t1)
[1] 4 4
> t1$X1 <- 1  
> t1$X2 <- 2
Note that R accepts this.  I think it's replaced t1$X1 with a 
vector of 1's using the usual repeating rules, and similarly for
X2.  So I should still have a 4 X 4 data frame.
R seems to think so too.
> dim(t1)  
[1] 4 4
> is.data.frame(t1)
[1] TRUE

But:
> print(t1)
Error: dim<- length of dims do not match the length of object
> t1[, 1:2]
$X1
[1] 1

$X2
[1] 2

whereas t1[,3:4] gives me a 4 x 2 data frame as I would expect.

It seems that the "listness" of t1 is beating out it's
"matrixness"
(having dimension and so on).  It is possible to get what I want with
t1$X1 <- rep(1, length(t1$X1)).

As I said before, I thought that t1$X1 <- 1 would replace the whole first
column of t1 with 1's.  Did other people think so too?  And is that in
fact the way R should work?

(This example used all numeric columns, in which case a matrix would work.
In some of my real applications, that's not necessarily practical.)

Thanks, 

Matt Wiener



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Matthew Wiener <mcw at ln.nimh.nih.gov> writes:
> t1 <- data.frame(matrix(rnorm(16), nc=4))
> t1$X1 <- 1  
> t1$X2 <- 2
> print(t1)
> Error: dim<- length of dims do not match the length of object
Well, it is prototype-compatible. Splus 5.3 does likewise. A way out
is

t1<-data.frame(unclass(t1))

However, we do seem to have a bug in the area:
>  t1 <- data.frame(matrix(rnorm(16), nc=4))
>  transform(t1,X1=1)
Error: dim<- length of dims do not match the length of
object
>  transform(t1,X1=1,X2=2)
  X1 X2         X3         X4
1  1  2  0.6020524  1.1595935
2  1  2 -1.8212204 -0.2118746
3  1  2  1.5506706  0.0680275
4  1  2 -0.8872958  0.4709870

This boils down to the construction

dfr[1]<-list(X1=1)

which doesn't extend dfr$X1, whereas

dfr[c(1,2)]<-list(X1=3,X2=4)

does. Splus handles both cases. 

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._