allow showProgress=INTEGER to set progress bar update time

NEWS.md

@@ -16,6 +16,8 @@
 
 1. `nafill()`, `setnafill()` extended to work on logical vectors (part of [#3992](https://github.com/Rdatatable/data.table/issues/3992)). Thanks @jangorecki for the request and @MichaelChirico for the PR.
 
+2. `[,showProgress=]` and `options(datatable.showProgress)` now accept an integer to control the progress bar update interval in seconds, allowing finer control over progress reporting frequency; `TRUE` uses the default 3-second interval, [#6514](https://github.com/Rdatatable/data.table/issues/6514). Thanks @ethanbsmith for the report and @ben-schwen for the PR.
+
 ### Notes
 
 1. {data.table} now depends on R 3.5.0 (2018).

R/data.table.R

@@ -244,7 +244,7 @@ replace_dot_alias = function(e) {
   if ((isTRUE(which)||is.na(which)) && !missing(j)) stopf("which==%s (meaning return row numbers) but j is also supplied. Either you need row numbers or the result of j, but only one type of result can be returned.", which)
   if (is.null(nomatch) && is.na(which)) stopf("which=NA with nomatch=0|NULL would always return an empty vector. Please change or remove either which or nomatch.")
   if (!with && missing(j)) stopf("j must be provided when with=FALSE")
-  if (!missing(by) && !isTRUEorFALSE(showProgress)) stopf("%s must be TRUE or FALSE", "showProgress")
+  if (!missing(by) && !(isTRUEorFALSE(showProgress) || (is.numeric(showProgress) && length(showProgress)==1L && showProgress >= 0))) stopf("showProgress must be TRUE, FALSE, or a single non-negative number") # nocov
   irows = NULL  # Meaning all rows. We avoid creating 1:nrow(x) for efficiency.
   notjoin = FALSE
   rightcols = leftcols = integer()
@@ -1972,7 +1972,7 @@ replace_dot_alias = function(e) {
     }
     ans = c(g, ans)
   } else {
-    ans = .Call(Cdogroups, x, xcols, groups, grpcols, jiscols, xjiscols, grporder, o__, f__, len__, jsub, SDenv, cols, newnames, !missing(on), verbose, showProgress)
+    ans = .Call(Cdogroups, x, xcols, groups, grpcols, jiscols, xjiscols, grporder, o__, f__, len__, jsub, SDenv, cols, newnames, !missing(on), verbose, as.integer(showProgress))
   }
   # unlock any locked data.table components of the answer, #4159
   # MAX_DEPTH prevents possible infinite recursion from truly recursive object, #4173

man/data.table.Rd

@@ -181,7 +181,7 @@ data.table(\dots, keep.rownames=FALSE, check.names=FALSE, key=NULL, stringsAsFac
 
   \item{env}{ List or an environment, passed to \code{\link{substitute2}} for substitution of parameters in \code{i}, \code{j} and \code{by} (or \code{keyby}). Use \code{verbose} to preview constructed expressions. For more details see \href{../doc/datatable-programming.html}{\code{vignette("datatable-programming")}}. }
 
-  \item{showProgress}{ \code{TRUE} shows progress indicator with estimated time to completion for lengthy "by" operations. }
+  \item{showProgress}{ \code{TRUE} (default when \code{interactive()}) shows a progress indicator with estimated time to completion for lengthy "by" operations, updating every 3 seconds. An integer value controls the update interval in seconds (minimum 3). \code{FALSE} disables the progress indicator. }
 }
 \details{
 \code{data.table} builds on base \R functionality to reduce 2 types of time:\cr

src/dogroups.c

@@ -86,9 +86,10 @@ SEXP dogroups(SEXP dt, SEXP dtcols, SEXP groups, SEXP grpcols, SEXP jiscols, SEX
   SEXP SDall = PROTECT(findVar(install(".SDall"), env)); nprotect++;  // PROTECT for rchk
   SEXP SD = PROTECT(findVar(install(".SD"), env)); nprotect++;
 
-  const bool showProgress = LOGICAL(showProgressArg)[0]==1 && ngrp > 1; // showProgress only if more than 1 group
+  int updateTime = INTEGER(showProgressArg)[0];
+  const bool showProgress = updateTime > 0 && ngrp > 1; // showProgress only if more than 1 group
   double startTime = (showProgress) ? wallclock() : 0; // For progress printing, startTime is set at the beginning
-  double nextTime = (showProgress) ? startTime+3 : 0; // wait 3 seconds before printing progress
+  double nextTime = (showProgress) ? startTime + MAX(updateTime, 3) : 0; // wait at least 3 seconds before starting to print progress
 
   hashtab * specials = hash_create(3 + ngrpcols + xlength(SDall)); // .I, .N, .GRP plus columns of .BY plus SDall
   PROTECT(specials->prot); nprotect++;
@@ -451,17 +452,17 @@ SEXP dogroups(SEXP dt, SEXP dtcols, SEXP groups, SEXP grpcols, SEXP jiscols, SEX
     // could potentially refactor to use fread's progress() function, however we would lose some information in favor of simplicity.
     double now;
     if (showProgress && (now=wallclock())>=nextTime) {
+      // # nocov start. Requires long-running test case
       double avgTimePerGroup = (now-startTime)/(i+1);
       int ETA = (int)(avgTimePerGroup*(ngrp-i-1));
       if (hasPrinted || ETA >= 0) {
-        // # nocov start. Requires long-running test case
         if (verbose && !hasPrinted) Rprintf(_("\n"));
         Rprintf("\r"); // # notranslate. \r is not internationalizable
         Rprintf(_("Processed %d groups out of %d. %.0f%% done. Time elapsed: %ds. ETA: %ds."), i+1, ngrp, 100.0*(i+1)/ngrp, (int)(now-startTime), ETA);
-        // # nocov end
       }
-      nextTime = now+1;
+      nextTime = now+updateTime;
       hasPrinted = true;
+      // # nocov end
     }
     ansloc += maxn;
     if (firstalloc) {