[R] `[.data.frame`(df3, , -2) and NA columns

[Tony Plate]

Assigning a name vector to a dataframe that is shorter than the number of 
columns results in some columns having NA values for their names.

"[.data.frame" has the following code in it:

         cols <- names(x)
         ...
         if (any(is.na(cols)))
             stop("undefined columns selected")

so, if a dataframe x has NA values for column names, you should expect 
x[...] to *sometimes* stop with that error (with a bit of reading and 
testing you could probably work out exactly when that error will occur).

-- Tony Plate

Dieter Menne wrote:
> Dear baseRs,
> 
> I recently made a mistake when renaming data frame columns, accidentally
> creating an NA column. I found the following strange behavior when negative
> indexes are used.
> 
> Can anyone explain what happens here.  No "workarounds" required, just  curious.
> 
> Dieter
> 
> Version: Windows, R version 2.6.1 (2007-11-26)
> 
> #-----------------------------
> df = data.frame(a=0:10,b=10:20)
> df[,-2] #ok
> names(df)=c("A") # implicitly creates an NA column
> df[,-2] 
> df[,-2,drop=FALSE]       # has nothing to do with drop
> 
> df3 = data.frame(a=0:10,b=10:20,c=20:30)
> df3[,-2] #ok
> names(df3)=c("A","B") #creates an NA column
> df3[,-2] # error
> # Error in `[.data.frame`(df3, , -2) : undefined columns selected
> 
> names(df3)[3]="NaN" # another reserved word
> df3[,-2] # no problem
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>