bug? rbindlist idcol constructed incorrectly

[franknarf1]

Looking at this SO question

# http://stackoverflow.com/questions/33661645/r-number-of-non-na-records-by-column-grouped
list_1 <- list(a = c(44,47), dens = c(2331,1644))
list_2 <- list(a=66, dens= 1890)
list_3 <- list(a=c(44,46,48,50), dens=c(8000,1452,1596,7521)) 
mylist <- list(list_1, list_2, list_3)
names(mylist)<-c("ID_1","ID_2","ID_3")

I thought I could use rbindlist, but

rbindlist(mylist, id=TRUE)

    .id  a dens
1: ID_1 44 2331
2: ID_1 47 1644
3: ID_2 66 1890
4: ID_2 44 8000
5: ID_3 46 1452
6: ID_3 48 1596
7: ID_1 50 7521
Warning message:
In set(ans, j = idcol, value = rep.int(nm, len)) :
  Supplied 6 items to be assigned to 7 items of column '.id' (recycled leaving remainder of 1 items).

As suggested by the warning, the final .id is wrong. I see the same warning with various subsets of mylist.

By the way, the "How to file a bug report" link is still dead over here: https://github.com/Rdatatable/data.table/blob/master/Contributing.md

[franknarf1]

Looking at the code for rbindlist, I think I found the problem:

vapply(mylist, NROW, 0L)
# 2 2 2

where we need 2 1 4. To get the latter, vapply(lapply(mylist, setDT), NROW, 0L) works, but alters the input. I'm sure you guys know a better way to do it.

[eantonya]

maybe vapply(l, function(x) length(x[[1]]), 0L)?

On Wed, Nov 11, 2015 at 5:52 PM, franknarf1 notifications@github.com
wrote:

Looking at the code for rbindlist, I think I found the problem:

vapply(mylist, NROW, 0L)

2 2 2

where we need 2 1 4. To get the latter, vapply(lapply(mylist, setDT),
nrow, 0L) works, but alters the input. I'm sure you guys know a better
way to do it.

—
Reply to this email directly or view it on GitHub
#1432 (comment)
.