by= ignored if there is no j expression

[st-pasha]

> DT = data.table(A=c(1,2,1,2,1,2), B=c(1,2,1,1,2,2))
> DT[, , by=A+B]  # result ignores the by= clause
   A B
1: 1 1
2: 2 2
3: 1 1
4: 2 1
5: 1 2
6: 2 2
> DT[, .(A, B), by=A+B]  # this is correct, DT[,,by=A+B] expected to be same as this
   A + B A B
1:     2 1 1
2:     2 1 1
3:     4 2 2
4:     4 2 2
5:     3 2 1
6:     3 1 2

[franknarf1]

Related #1269 (comment) and later comments

[MichaelChirico]

Not sure I follow your reasoning for your second example... is that for convenience? easy to imagine I have some complicated expression by = f(x1, .., xn) and don't really care about the inputs to f.

[st-pasha]

One could argue whether DT[,,B] makes sense or not.
If it doesn't, then an error should be thrown.
If it does, then the result should be the DT reordered in such way that the values in column B are grouped together. And we should also decide whether B ought to be moved to the front or kept in its place, or both.

However, I don't think silently ignoring the by argument is the right thing to do...

[MichaelChirico]

definitely agreed, and in fact I think this is natural syntax for just returning the groups. but in that case this issue is a duplicate of the issue Frank referenced IIUC. the novelty here was your second example which I don't quite follow

…

On Thu, Jan 10, 2019, 8:28 AM Pasha Stetsenko ***@***.*** wrote: One could argue whether DT[,,B] makes sense or not. If it doesn't, then an error should be thrown. If it does, then the result should be the DT reordered in such way that the values in column B are grouped together. And we should also decide whether B ought to be moved to the front or kept in its place, or both. However, I don't think silently ignoring the by argument is the right thing to do... — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#3263 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHQQdSw72is90-yCUv7J8_SPbfL5IQejks5vBolKgaJpZM4Z4JJD> .

[st-pasha]

But the second example already does the right thing: DT[, .(A,B), by=A+B] means group the data.table by expression A+B, then select columns A and B. And that's exactly what it does, also adding the group-by column A+B at the front.

[mattdowle]

Please use dev not release when reporting bugs.

> DT = data.table(A=c(1,2,1,2,1,2), B=c(1,2,1,1,2,2))
> DT[, , by=A+B] 
...
Warning message:
In `[.data.table`(DT, , , by = A + B) :
  i and j are both missing so ignoring the other arguments
>

[MichaelChirico]

@st-pasha sorry I see the confusion, I was referring to your comment:

this is correct, DT[,,by=A+B] expected to be same as this

[mattdowle]

Closing because the warning is there and that was the focus of this issue.
Can follow up in #3262 perhaps.