(PR#1608) merge.data.frame can coerce character vectors to factor in some circumstances (PR#1608)

If the following two conditions are met:

1) all.x is TRUE

2) at least 1 row in y does not have a match in x

then any character vectors in y will be coerced to be factors. Here is a simple
example (previously provided on r-devel):

x <- data.frame(a = 1:4)
y <- data.frame(b = LETTERS[1:3])
y$b <- as.character(y$b)
z <- merge(x, y, by = 0, all.x = TRUE)
z

  Row.names a    b
1         1 1    A
2         2 2    B
3         3 3    C
4         4 4   <NA>

sapply(z, data.class)

Row.names         a         b
 "factor" "numeric"  "factor"

This problem could be fixed by changing the line in merge.data.frame:

for (i in seq(along = y)) is.na(y[[i]]) <- (lxy + 1):(lxy + nxx)

to:

for (i in seq(along = y)) y[((lxy + 1):(lxy + nxx)), i] <- NA

To the extent that this is a feature rather than a bug (if so, I would like to
know why),

 then I would suggest that the following sentence be added to the
documentation for merge at the end of the section on all.x

"Be aware that, if all.x equals `TRUE', character vectors in `y' will be
converted to factors if any rows in y have no matching row in `x'."