Data.table: La clé casse "par" la fonctionnalité

Je ne suis pas tout à fait certain de ce qui cause cela, alors voici le WE le plus minimal que j'ai pu trouver

library(data.table) # Tested on v 1.9.7
dt <-  data.table( origin = c("A", "A", "A", "A", "A", "A", "B", "B", "A", "A", "C", "C", "B", "B", "B", "B", "B", "C", "C", "B", "A", "C", "C", "C", "C", "C", "A", "A", "C", "C", "B", "B"),
                   destination = c("A", "A", "A", "A", "B", "B", "A", "A", "C", "C", "A", "A", "B", "B", "B", "C", "C", "B", "B", "A", "B", "C", "C", "C", "A", "A", "C", "C", "B", "B", "C", "C"),
                   points_in_dest = c(5, 5, 5, 5, 4, 4, 5, 5, 3, 3, 5, 5, 4, 4, 4, 3, 3, 4, 4, 5, 4, 3, 3, 3, 5,5, 3, 3, 4, 4, 3, 3),
                   depart_time = c(7, 8, 16, 18, 7, 8, 16, 18, 7, 8, 16, 18, 7, 8, 16, 7, 8, 16, 18, 8, 16, 7, 8, 18, 7, 8, 16, 18, 7, 8, 16, 18),   
                   travel_time = c(0, 0, 0, 0, 70, 10, 70, 10, 10, 10, 70, 70, 0, 0, 0, 70, 10, 10, 70, 70, 10, 0, 0, 0, 10, 70, 10, 70, 10, 70, 70, 10) )

dt[ depart_time<=8  & travel_time < 60, condition1 := TRUE]
dt[ depart_time>=16 & travel_time < 60, condition2 := TRUE] 

setkey(dt, origin, destination)
res <- unique(dt[(condition1)])[unique(dt[(condition2)]), 
                                on = c(destination = "origin", origin = "destination"), 
                                nomatch = 0L]
res[, .(points = sum(points_in_dest)),  keyby = origin]
#    origin points
#1:      A      5
#2:      A      4
#3:      B      4
#4:      B      3
#5:      C      5
#6:      C      4
#7:      C      3

Comme vous pouvez le voir, by n'a pas fonctionné comme prévu et toutes les lignes ont été renvoyées. Il s'agit évidemment d'un problème de saisie car ce qui suit résout ce problème

setattr(res, "sorted", NULL)
res[, .(points = sum(points_in_dest)), keyby = origin]
#    origin points
#1:      A      9
#2:      B      7
#3:      C     12

Ou, alternativement, pré-classer origin à un facteur

res[, .(points = sum(points_in_dest)), keyby = factor(origin)]
#    factor points
#1:      A      9
#2:      B      7
#3:      C     12

Ceci a été tiré de cette question SO http://stackoverflow.com/questions/37239649/aggregate-data-table-based-on-condition-in-another-row