Data tables won't merge if by argument has names

[sebastian-c]

I've been trying the merge some data tables, but it seems that if the 'by' argument has names, they won't merge, giving an error:

Error in forderv(x, by = rightcols) : 'by' value -2147483648 out of range [1,8]

I've managed to make as small an example as I can to reproduce the error:

tab1 <- structure(list(a = c(1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 
3L, 3L, 3L, 3L, 3L, NA), b = c(2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 
2, 2, 2, 2, 2, NA), c = c(2010L, 2011L, 2012L, 2013L, 2014L, 
2010L, 2011L, 2012L, 2013L, 2014L, 2010L, 2011L, 2012L, 2013L, 
2014L, NA), d = c(-0.625996555127917, 0.325673822554901, 1.68461340609045, 
-2.09358233735785, -0.444517113337327, 0.247434396526436, -0.595608214998598, 
0.779085729216547, 0.343640797995949, -1.03199833192338, 0.499177323692263, 
0.142404120359858, 0.124892949669588, 0.28962454726758, 1.45581442296567, 
1.27173157262308), e = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA), f = c(NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA), g = c(NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA), h = c(1.82940602609317, 
0.337391723803077, 1.49770018138232, -0.353146958400835, 1.09995073678083, 
-0.578792240046591, 1.55076285478448, 0.919702101279719, 1.91852234353312, 
0.902454834121264, -1.37138551168091, 0.416357331689734, 0.306641948600609, 
-1.1230027058004, -0.621123012856139, 0.645928414645345)), .Names = c("a", 
"b", "c", "d", "e", "f", "g", "h"), row.names = c(NA, -16L), class = "data.frame")

tab2 <- structure(list(a = c(1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 
3L, 3L, 3L, 3L, 3L, NA), b = c(2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 
2, 2, 2, 2, 2, NA), c = c(2010L, 2011L, 2012L, 2013L, 2014L, 
2010L, 2011L, 2012L, 2013L, 2014L, 2010L, 2011L, 2012L, 2013L, 
2014L, NA), D = structure(c(1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 
2L, 1L, 1L, 1L, 1L, 2L, NA), .Label = c("", "M"), class = "factor"), 
    E = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
    NA, NA, NA), F = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
    NA, NA, NA, NA, NA, NA), G = c(NA, NA, NA, NA, NA, NA, NA, 
    NA, NA, NA, NA, NA, NA, NA, NA, NA), H = structure(c(1L, 
    1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 2L, NA
    ), .Label = c("", "M"), class = "factor")), .Names = c("a", 
"b", "c", "D", "E", "F", "G", "H"), row.names = c(NA, -16L), class = "data.frame")

tab1 <- data.table(tab1)
tab2 <- data.table(tab2)

keys <- c("a", "b", "c")

merge(tab1, tab2, by=keys)
names(keys) <- letters[24:26]

merge(tab1, tab2, by=keys)

Notice that the first merge works, but the second merge gives the aforementioned error.

[franknarf1]

If you instead use

names(keys) <- keys

the merge goes through fine. Presumably, the the names are used in some (as yet undocumented) way. You could mess around with permutations, like

names(keys) <- rev(keys)

to uncover what's going on.

[arunsrinivasan]

Thanks for the nice reprex @sebastian-c, should be fixed now. Please verify.

[sebastian-c]

Confirmed, the bug has been fixed.