Integer64 columns not assigned NA when doing a left merge

[abielr]

Handle NA integer64 properly in:

• merge
• rbindlist (following @jangorecki's comment below)

require(data.table)
require(bit64)
x1 <- data.table(id=1, value=as.integer64(1))
x2 <- data.table(id=c(1,2))
merge(x2, x1, by="id", all.x=T)

Yields

#    id               value
#1:  1                   1
#2:  2 9218868437227407266

This is with data.table 1.9.6.

[dlithio]

I think this is a duplicate.

[arunsrinivasan]

Thanks for linking to the old issue @dlithio. Maybe we can fix it internally..

[jangorecki]

The same happens when rbindlist(fill=TRUE). This probably can be fixed in the same place as merge so I'm not creating new issue for that, let me know if you want separate one for rbindlist fill.

library(data.table)
library(bit64)
rbindlist(list(
    data.table(a=1:2, b=as.integer64(c(1,NA))),
    data.table(a=3L)
), fill=TRUE)
#   a                   b
#1: 1                   1
#2: 2                  NA
#3: 3 9218868437227407266

[andrewkho]

I'm not sure if I should create a new issue for this, but I am having this problem when using the := operator, but not when using merge.

> packageVersion("data.table")
[1] ‘1.10.4’
> packageVersion("bit64")
[1] ‘0.9.7’
> dt1 <- data.table(id=c(1,2), x=bit64::as.integer64(c(100,200)))
> dt2 <- data.table(id=c(1,3))
> dt1
   id   x
1:  1 100
2:  2 200
> dt2
   id
1:  1
2:  3
> merge(dt1,dt2, all=T)
   id   x
1:  1 100
2:  2 200
3:  3  NA
> dt2[J(dt1), x := i.x, on=.(id)]
> dt2
   id                   x
1:  1                 100
2:  3 9218868437227407266

Update: I guess this is the expected behaviour:

> merge(dt1, dt2, all.y = T)
   id   x
1:  1 100
2:  3  NA

I have a lot of code using the := method, and I'm not sure if it's feasible to switch everything to use merge instead

[MichaelChirico]

confirming problem for current dev

…

On Tue, Oct 3, 2017 at 5:47 AM, andrewkho ***@***.***> wrote: I'm not sure if I should create a new issue for this, but I am having this problem when using the := operator, but not when using merge. > packageVersion("data.table") [1] ‘1.10.4’ > packageVersion("bit64") [1] ‘0.9.7’ > dt1 <- data.table(id=c(1,2), x=bit64::as.integer64(100,200)) > dt2 <- data.table(id=c(1,3)) > dt1 id x 1: 1 100 2: 2 100 > dt2 id 1: 1 2: 3 > merge(dt1,dt2, all=T) id x 1: 1 100 2: 2 100 3: 3 NA > dt2[J(dt1), x := i.x, on=.(id)] > dt2 id x 1: 1 100 2: 3 9218868437227407266 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1459 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHQQdRNN5xLwJ4C3TIc-jV5zJrfXo4RQks5soVoEgaJpZM4Gul8T> .