rbindlist use.names=TRUE returns random column order when ncol>255

[mattdowle]

Follow up to #3306. The printing issue there is fixed by not using use.names=TRUE in that case. But rbindlist in general still has the issue. Also see: #3352 (comment)

minimal example:

ncol = 30
DT = setDT(replicate(ncol, rnorm(3L), simplify = FALSE))
identical(colnames(rbind(DT[1], DT[3])), colnames(DT))
# [1] TRUE


ncol = 300
DT = setDT(replicate(ncol, rnorm(3L), simplify = FALSE))
identical(colnames(rbind(DT[1], DT[3])), colnames(DT))
# [1] FALSE

[ChristK]

Possibly related...
rbind changes column order when there are duplicate column names

require(data.table) #v1.12.1 built 2019-02-08 09:09:06 UTC;
dt1 <- data.table(a=1, b=2, c=3, b=4)
dt2 <- data.table(a=1, b=2, c=3, b=4)
rbind(dt1, dt2, use.names = TRUE) # use.names = FALSE does not reorder the columns
# a b b c
# 1: 1 2 4 3
# 2: 1 2 4 3

[MichaelChirico]

yes it'll be the same issue

…

On Sat, Feb 9, 2019, 5:23 AM Chris Kypridemos ***@***.*** wrote: Possibly related... rbind changes column order when there are duplicate column names require(data.table) #v1.12.1 built 2019-02-08 09:09:06 UTC;dt1 <- data.table(a=1, b=2, c=3, b=4)dt2 <- data.table(a=1, b=2, c=3, b=4) rbind(dt1, dt2, use.names = TRUE) # use.names = FALSE does not reorder the columns# a b b c# 1: 1 2 4 3# 2: 1 2 4 3 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#3373 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHQQdUIDORlu947IBa9PfblPOTwcOX19ks5vLerIgaJpZM4auWYF> .