[Rd] as.data.frame: Error in "names<-.default" (PR#7808)

[Liaw, Andy]

> From: Robert.McGehee at geodecapital.com
> 
> Hello,
> I found a potential problem in R 2.1.0 (and R 2.0.1)
> 
> I expect that
> 
> > tmp <- FUN(x1, x2, x3, x4)
> > as.data.frame(tmp)
> 
> is the same as
> > as.data.frame(FUN(x1, x2, x3, x4))
> 
> since the tmp variable in this case is unnecessary.
> 
> However, below I will demonstrate that under an odd set of 
> conditions, I
> can correctly perform as.data.frame(tmp), but not 
> as.data.frame(FUN(x1,
> x2, x3, x4)). 
> 
> ## This code works correctly
> FUN <- function(x1, x2, x3, x4) 
>      cbind(x1[, 1, 1:2], x1[, 2, 1:2])[, 1]
> 
> x1 <- array(1:9, c(3, 3, 3))
> tmp <- FUN(x1[1:3, , ], x2 = c("a", "b"), x3 = c("a", "b"), 
> x4 = c("a",
> "b"))
> 
> ## Works correctly
> as.data.frame(tmp)  
> 
>   tmp
> 1   1
> 2   2
> 3   3
> 
> 
> ## This (supposedly equivalent) code gives an error
> 
> as.data.frame(FUN(x1[1:3,,], x2 = c("a", "b"), x3 = c("a", "b"), x4 =
> c("a", "b")))
> 
> Error in "names<-.default"(`*tmp*`, value = c("FUN(x1[1:3, , ], x2 =
> c(\"a\", \"b\"), x3 = c(\"a\", \"b\"), x4 = c(\"a\", ",  : 
> 	'names' attribute [2] must be the same length as the vector [1]
> 
> Note, that while the extra (unused) arguments in FUN seem unnecessary,
> as well as the odd indexing, the problem disappears when I remove the
> extraneous values. Unfortunately, I have not found a more 
> elegant way to
> present this problem, but hopefully this code will be helpful.

The basic problem, I think, boils down to something like this:

> f <- function(x) deparse(substitute(x))
> f(FUN(x1[1:3,,], x2=c("a","b"), x3=c("a", "b"), x4=c("a", "b")))
[1] "FUN(x1[1:3, , ], x2 = c(\"a\", \"b\"), x3 = c(\"a\", \"b\"), x4 =
c(\"a\", "
[2] "    \"b\"))"


which is caused by deparse() chopping up the expression.  The fix would be
to set the width.cutoff argument to something large.  Here's a proposed
patch:

--- R-2.1.0/src/library/base/R/dataframe.R      2005-04-18
06:19:15.000000000 -0
400
+++ R-2.1.0-fix/src/library/base/R/dataframe.R  2005-04-22
23:17:18.972665600 -0
400
@@ -114,7 +114,7 @@
 as.data.frame.vector <- function(x, row.names = NULL, optional = FALSE)
 {
     nrows <- length(x)
-    nm <- deparse(substitute(x))
+    nm <- deparse(substitute(x), width.cutoff=500)
     if(is.null(row.names)) {
        if (nrows == 0)
            row.names <- character(0)
@@ -235,7 +235,7 @@
        as.data.frame.model.matrix(x, row.names, optional)
     else { # as.data.frame.vector without removing names
         nrows <- length(x)
-        nm <- deparse(substitute(x))
+        nm <- deparse(substitute(x), width.cutoff=500)
         if(is.null(row.names)) {
             if (nrows == 0)
                 row.names <- character(0)

(I used width.cutoff=500, as ?deparse says that the max.  I'd imagine the
number of characters allowed for valid symbol names in R is probably lower
than that?)

Andy
 
> Robert
> 
> Robert McGehee
> Geode Capital Management, LLC
> 53 State Street, 5th Floor | Boston, MA | 02109
> Tel: 617/392-8396    Fax:617/476-6389
> mailto:robert.mcgehee at geodecapital.com
> 
> 
> 
> This e-mail, and any attachments hereto, are intended for 
> us...{{dropped}}
> 
> ______________________________________________
> R-devel at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
> 
> 
>