corrected

Author: venom1204

NEWS.md

@@ -107,7 +107,7 @@
 
 18. `fwrite` now allows `dec` to be the same as `sep` for edge cases where only one will be written, e.g. 0-row or 1-column tables. [#7227](https://github.com/Rdatatable/data.table/issues/7227). Thanks @MichaelChirico for the report and @venom1204 for the fix.
 
-22. Using `by=` or `keyby=` with a simple numeric or character vector in `j` (e.g. `DT[, 1:2, by=grp]`) used to silently ignore the grouping argument. This now issues a warning to alert the user that grouping is not applied in this syntax and guides them to use the `.SD` idiom instead. [#5397](https://github.com/Rdatatable/data.table/issues/5397). Thanks to @mcol for the report and @venom1204 for the fix.
+19. Using `by=` or `keyby=` with a simple numeric or character vector in `j` (e.g. `DT[, 1:2, by=grp]`) used to silently ignore the grouping argument. This now issues a warning to alert the user that grouping is not applied in this syntax and guides them to use the `.SD` idiom instead. [#5397](https://github.com/Rdatatable/data.table/issues/5397). Thanks to @mcol for the report and @venom1204 for the fix.
 
     ```r
     DT = data.table(a=1:4, grp=c(1,1,2,2))

R/data.table.R

@@ -739,11 +739,11 @@ replace_dot_alias = function(e) {
       if (!length(j) && !notj) return( null.data.table() )
       if (is.factor(j)) j = as.character(j)  # fix for FR: #358
       if (is.character(j)) {
-        if (!missingby && (missing(with) || isTRUE(with))) {
-            warning(
-                "`by` or `keyby` is ignored when `j` is a character vector used for column selection. ",
-                "Perhaps you intended to use `.SD`? For example: DT[, .SD[, ", deparse(jsub), "], by = ...]"
-            )
+        if (!missingby) {
+        warning(
+            "`by` or `keyby` is ignored when `j` is a character vector used for column selection. ",
+            "Perhaps you intended to use `.SD`? For example: DT[, .SD[, ", deparse(jsub), "], by = ...]"
+        )
         }
         if (notj) {
           if (anyNA(idx <- chmatch(j, names_x)))
@@ -768,9 +768,11 @@ replace_dot_alias = function(e) {
         # else the NA in ansvals are for join inherited scope (test 1973), and NA could be in irows from join and data in i should be returned (test 1977)
         #   in both cases leave to the R-level subsetting of i and x together further below
       } else if (is.numeric(j)) {
-        if (!missingby && (missing(with) || isTRUE(with))) {
-            warning(
-                "`by` or `keyby` is ignored when `j` is a numeric vector used for column selection. ", "Perhaps you intended to use `.SD`? For example: DT[, .SD[, ", deparse(jsub), "], by = ...]")
+        if (!missingby) {
+        warning(
+            "`by` or `keyby` is ignored when `j` is a numeric vector used for column selection. ",
+            "Perhaps you intended to use `.SD`? For example: DT[, .SD[, ", deparse(jsub), "], by = ...]"
+        )
         }
         j = as.integer(j)
         if (any(w <- (j>ncol(x)))) stopf("Item %d of j is %d which is outside the column number range [1,ncol=%d]", idx <- which.first(w), j[idx], ncol(x))

inst/tests/tests.Rraw

@@ -21624,7 +21624,7 @@ local({
 DT = data.table(a=1:4, b=5:8, g=c(1,1,2,2))
 test(2339.1, DT[, 1:2, by=g], DT[, 1:2], warning="`by` or `keyby` is ignored")
 test(2339.2, DT[, 2:1, keyby=g], DT[, 2:1], warning="`by` or `keyby` is ignored")
-test(2339.3, DT[, c("b", "a"), by=g, with=FALSE], DT[, c("b", "a")])
+test(2339.3, DT[, c("b", "a"), by=g, with=FALSE], DT[, c("b", "a")], warning="`by` or `keyby` is ignored")
 expected_sd = data.table(g=c(1,1,2,2), a=1:4, b=5:8)
 test(2339.4, DT[, .SD[, 1:2], by=g], expected_sd)
 expected_single_int = data.table(g=c(1,2), V1=c(1,1))

man/data.table.Rd

@@ -97,7 +97,7 @@ data.table(\dots, keep.rownames=FALSE, check.names=FALSE, key=NULL, stringsAsFac
 
     See \href{../doc/datatable-intro.html}{\code{vignette("datatable-intro")}} and \code{example(data.table)}.}
 
-    \item{by}{ Column names are seen as if they are variables (as in \code{j} when \code{with=TRUE}). \emph{Note that `by` and `keyby` are ignored when `j` is a character or numeric vector used for selecting columns (i.e., when the internal `with=FALSE` is triggered).} The \code{data.table} is then grouped by the \code{by} and \code{j} is evaluated within each group. The order of the rows within each group is preserved, as is the order of the groups. \code{by} accepts:
+    \item{by}{ Column names are seen as if they are variables (as in \code{j} when \code{with=TRUE}). \emph{Note that \code{by} and \code{keyby} are ignored when \code{j} is a character or numeric vector used for selecting columns (i.e., when the internal \code{with=FALSE} is triggered).} The \code{data.table} is then grouped by the \code{by} and \code{j} is evaluated within each group. ...}    
     
     \itemize{
         \item A single unquoted column name: e.g., \code{DT[, .(sa=sum(a)), by=x]}