[datatable-help] Inconsistent [<- behavior

[Matthew Dowle]

Forgot to say, please add a bug.report(package="data.table") for the errors. 
Thanks.

"Matthew Dowle" <mdowle at mdowle.plus.com> wrote in message 
news:j4n44f$36p$1 at dough.gmane.org...
>
> Hi. Thanks for reporting. The 'correct' ways that work using := are the 
> best ways going forward. But agreed, will fix so that all ways work.
>
> "Leon Baum" <leonbaum2 at gmail.com> wrote in message 
> news:87sjo1nwnt.fsf at gmail.com...
>>
>> Hello,
>>
>> I believe that all of the following should result in the same data.table
>> (ignoring different orderings), but some of them give me errors:
>>
>>
>>> dt <- data.table(x=letters[1:2], y=rnorm(4))
>>> dt[c(1, 3), ]$y <- 0
>>> dt
>>     x         y
>> [1,] a  0.000000
>> [2,] b -1.645834
>> [3,] a  0.000000
>> [4,] b -1.311856
>> # correct
>>
>>> dt <- data.table(x=letters[1:2], y=rnorm(4))
>>> dt[x == "a", ]$y <- 0
>> Error in `[<-.data.frame`(x, i, j, value) : object 'x' not found
>> # bad
>>
>>> dt <- data.table(x=letters[1:2], y=rnorm(4))
>>> dt[x == "a", y:=0]
>>     x          y
>> [1,] a  0.0000000
>> [2,] b -0.1028541
>> [3,] a  0.0000000
>> [4,] b  0.3569112
>> # correct
>>
>>> dt <- data.table(x=letters[1:2], y=rnorm(4), key="x")
>>> dt["a", ]$y <- 0
>> Warning messages:
>> 1: In `[<-.data.frame`(x, i, j, value) :
>>  replacement element 1 has 2 rows to replace 1 rows
>> 2: In `[<-.data.frame`(x, i, j, value) :
>>  replacement element 2 has 2 rows to replace 1 rows
>>> dt
>>     x          y
>> [1,] a -0.4444458
>> [2,] a -2.1279203
>> [3,] b -1.9065767
>> [4,] b  0.5903236
>> [5,] a  0.0000000
>> # bad
>>
>>> dt <- data.table(x=letters[1:2], y=rnorm(4), key="x")
>>> dt["a", y:=0]
>>     x           y
>> [1,] a  0.00000000
>> [2,] a  0.00000000
>> [3,] b  1.05028739
>> [4,] b -0.02362426
>> # correct
>>
>> I am using R-2.13.1 and data.table 1.6.6.
>>
>> -Leon