Unclear error when calling rbind.ffdf() on ffdf with different number of columns

[nalimilan]

library(ffbase)
a <- as.ffd(data.frame(a=1:10, b=1:10))
b <- as.ffdf(data.frame(a=1:10, b=1:10, c=1:10))

rbind(a, b)
Error in data.frame(x.vmode = vmode(x), y.vmode = vmode(y), stringsAsFactors = FALSE) : 
  arguments imply different number of rows : 2, 3
In addition: Warning:
In ffdfappend(x, l) : names not identical when appending through ffdfappend

This error message is really misleading as it mentions rows while the number of columns differs between the two ffdf.

[nalimilan]

Sorry, this does not seem to be completely fixed. With a slightly different test case, I still get a weird error:

> iris1 <- as.ffdf(iris)
> iris2 <- as.ffdf(iris)
> rbind(iris1, subset(iris2, select=Species))
Erreur dans .subset2(dummy, i, exact = exact) : index out of bounds
> traceback()
9: `[[.ffdf`(dat, i)
8: dat[[i]]
7: lapply(list(...), function(x) {
       if (is.factor(x)) 
           levels(x)
       else x
   })
6: unlist(lapply(list(...), function(x) {
       if (is.factor(x)) 
           levels(x)
       else x
   }))
5: unique(unlist(lapply(list(...), function(x) {
       if (is.factor(x)) 
           levels(x)
       else x
   })))
4: appendLevels(levels(x[[i]]), dat[[i]])
3: ffdfappend(x, l)
2: rbind(deparse.level, ...)
1: rbind(iris1, subset(iris2, select = Species))

> ffdfappend(iris1, subset(iris2, select=Species))
Erreur dans .subset2(dummy, i, exact = exact) : index out of bounds

Also, with the test case of the original description, there are additional warnings that are not really useful:

a <- as.ffdf(data.frame(a=1:10, b=1:10))
b <- as.ffdf(data.frame(a=1:10, b=1:10, c=1:10))
> rbind(a, b)
Erreur dans ffdfappend(x, l) : Number of columns does not match
In addition: Warnings
1: In fc | sapply(physical(x), is.factor.ff) :
  size of a longer object is not a multiple of that of a shorter object [approximate translation]