> ff <- function() paste0(sample(letters, sample(5:8, 1), TRUE), collapse="")
> ll <- unique(replicate(1e4, ff()))
> DT <- data.table(v1=sample(ll, 1e6, TRUE), v2=sample(1:1e4, 1e6, TRUE), v3=sample(1:1e4, 1e6, TRUE))
>
> # add functions
> eddi <- function(dt=copy(DT)) {
+   dt[, list(v2=v2[v2 == v2[1]]), by = v1]
+ }
>
> andrey <- function(dt=copy(DT)) {
+   dt[, .SD[v2 == v2[1],], by = v1]
+ }
>
> arun <- function(dt=copy(DT)) {
+   tmp = dt[dt[, list(I=.I[1]), by=list(v1)]$I]
+   setkey(dt)[tmp]
+ }
>
> # benchmark
> microbenchmark(a1 <- eddi(), a2 <- andrey(), a3 <- arun(), times=2)
Unit: milliseconds
           expr  min   lq median   uq  max neval
   a1 <- eddi()  212  212    229  246  246     2
 a2 <- andrey() 4234 4234   4258 4282 4282     2
   a3 <- arun()  529  529    551  574  574     2
>
> setkey(a3, NULL)
             v1   v2   v3
    1: ufhjypjc 2260 3859
    2:  wnraxyc  867 9698
    3: ppvlahgr 6508 4440
    4: eaveaoey 4396 5581
    5:  mfoifvd 2005 5205
   ---
 9996:    rjrxx 2323 7954
 9997:  hbymqhz 3723 2110
 9998: kpiakpqh 1508 7807
 9999:    efqek 6592 5396
10000: assytuyp 5131 9038
>
> identical(a1, a2)
[1] FALSE
> identical(a1, a3)
[1] FALSE
> identical(a2, a3)
[1] FALSE
> all.equal(a2,a3)
[1] "Component 1: Lengths (10094, 10000) differ (string compare on first 10000)"
[2] "Component 1: 9740 string mismatches"
[3] "Component 2: Numeric: lengths (10094, 10000) differ"
[4] "Component 3: Numeric: lengths (10094, 10000) differ"