v2.7.2: address CRAN reviewer (Benjamin Altmann) feedback

In-cycle resubmission addressing all four items raised on the v2.7.1
review of 2026-04-27:

* DESCRIPTION: add methods references per CRAN cookbook --
  Breiman (2001) <doi:10.1023/A:1010933404324> and Ishwaran et al.
  (2008) <doi:10.1214/08-AOAS169>.

* shift() Rd: drop the public help page entirely. shift() is an
  internal numeric-vector lead/lag utility; its example used
  ggRandomForests:::shift(...). Marked the function @nord so no Rd
  is emitted; deleted man/shift.Rd.

* surv_partial.rfsrc(): replace cat("partial plot for: ...", "\n")
  with message(). Output is now suppressible via suppressMessages()
  and plays nicely inside notebooks / Shiny / quarto. Test updated:
  expect_output -> expect_message.

* surv_partial.rfsrc Rd example: wrap par(mfrow = c(2, 2)) with
  oldpar <- par(no.readonly = TRUE); on.exit(par(oldpar)) so the
  user's graphical parameters are restored on exit.

DESCRIPTION bumped to 2.7.2; NEWS gets a v2.7.2 heading enumerating
the four reviewer-driven fixes; cran-comments.md rewritten as an
in-cycle resubmission acknowledging Altmann's review.

Local R CMD check --as-cran on the v2.7.2 build: Status 2 NOTEs
(both unavoidable: CRAN incoming-feasibility + local NTP timestamp).
Examples [14s/14s] OK, donttest [17s/18s] OK, PDF manual OK, tests
OK.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: ehrlinger

DESCRIPTION

@@ -1,17 +1,22 @@
 Package: ggRandomForests
 Type: Package
 Title: Visually Exploring Random Forests
-Version: 2.7.1
+Version: 2.7.2
 Date: 2026-04-27
-Authors@R: person("John", "Ehrlinger",  
-  role = c("aut", "cre"), 
+Authors@R: person("John", "Ehrlinger",
+  role = c("aut", "cre"),
   email = "john.ehrlinger@gmail.com")
 License: MIT + file LICENSE
 Encoding: UTF-8
 Language: en-US
 URL: https://github.com/ehrlinger/ggRandomForests, https://ehrlinger.github.io/ggRandomForests/
 BugReports: https://github.com/ehrlinger/ggRandomForests/issues
-Description: Graphic elements for exploring Random Forests using the 'randomForest' or 'randomForestSRC' package for survival, regression and classification forests and 'ggplot2' package plotting.
+Description: Graphic elements for exploring Random Forests using the
+    'randomForest' or 'randomForestSRC' package for survival, regression
+    and classification forests and 'ggplot2' package plotting. Implements
+    visualisations of the methods described in Breiman (2001)
+    <doi:10.1023/A:1010933404324> and Ishwaran, Kogalur, Blackstone, and
+    Lauer (2008) <doi:10.1214/08-AOAS169>.
 Depends:
     R (>= 4.4.0),
     randomForestSRC (>= 3.4.0),

NEWS.md

@@ -1,5 +1,22 @@
 Package: ggRandomForests
-Version: 2.7.1
+Version: 2.7.2
+
+ggRandomForests v2.7.2
+=====================
+* Address CRAN reviewer (Benjamin Altmann) feedback on the v2.7.1
+  resubmission:
+  - Add methods references to `DESCRIPTION` (Breiman 2001 and
+    Ishwaran et al. 2008, with `<doi:...>` auto-links) per CRAN
+    cookbook.
+  - Drop the `man/shift.Rd` Rd file: `shift()` is an internal utility
+    and the example used `ggRandomForests:::shift(...)`. Marked the
+    function `@noRd` so it no longer generates a help page.
+  - Replace `cat()` in `surv_partial.rfsrc()` with `message()` so
+    progress output is suppressible (`suppressMessages()`) and plays
+    nicely inside notebooks / Shiny / quarto.
+  - Restore the user's `par()` settings in the
+    `surv_partial.rfsrc()` example via
+    `oldpar <- par(no.readonly = TRUE); on.exit(par(oldpar))`.
 
 ggRandomForests v2.7.1
 =====================

R/surv_partial.rfsrc.R

@@ -99,7 +99,10 @@
 #' pdta1 <- get.partial.plot.data(partial.obj, target = 1)
 #' pdta2 <- get.partial.plot.data(partial.obj, target = 2)
 #'
-#' par(mfrow=c(2,2))
+#' # Save and restore the user's graphical parameters per CRAN policy.
+#' oldpar <- par(no.readonly = TRUE)
+#' on.exit(par(oldpar))
+#' par(mfrow = c(2, 2))
 #' plot(lowess(pdta1$x, pdta1$yhat),
 #'      type = "l", xlab = "age", ylab = "adjusted years lost relapse")
 #' plot(lowess(pdta2$x, pdta2$yhat),
@@ -132,8 +135,9 @@ surv_partial.rfsrc <- function(rforest, var_list, npts = 25, partial.type = "sur
   )
   ###----------Partial dependency estimation, for each variable, at each time point ----
   surv.lst <- lapply(var_list, function(xvar) {
-    ## extract the key variable
-    cat("partial plot for:", xvar, "\n")
+    ## extract the key variable. Use message() (suppressible) instead of
+    ## cat() so the function plays nicely inside notebooks/Shiny/quarto.
+    message("partial plot for: ", xvar)
 
     ## determine the partial plot data
     xv <- sort(unique(rforest$xvar[, xvar]))

R/utils.R

@@ -15,27 +15,17 @@
 ## None of these are exported to end-users.
 
 # --------------------------------------------------------------------------- #
-#' Lead / lag shift for numeric vectors
-#'
-#' @param x a numeric vector of values
-#' @param shift_by an integer of length 1, giving the number of positions
-#'   to lead (positive) or lag (negative) by
-#'
-#' @details Lead and lag are useful for comparing values offset by a constant
-#' (e.g. the previous or next value).
-#'
-#' Taken from:
-#' http://ctszkin.com/2012/03/11/generating-a-laglead-variables/
-#'
-#' This function allows removal of the dplyr::lead dependency.
-#'
-#' @keywords internal
-#' @examples
-#' d <- data.frame(x = 1:15)
-#' # generate lead variable
-#' d$df_lead2 <- ggRandomForests:::shift(d$x, 2)
-#' # generate lag variable
-#' d$df_lag2 <- ggRandomForests:::shift(d$x, -2)
+# Internal: lead / lag shift for numeric vectors.
+#
+# `x`        numeric vector of values.
+# `shift_by` integer length 1 giving the number of positions to lead
+#            (positive) or lag (negative) by; can also be a vector to
+#            return a matrix of shifts.
+#
+# Removes the dplyr::lead dependency.  Adapted from
+# http://ctszkin.com/2012/03/11/generating-a-laglead-variables/
+#
+# @noRd
 shift <- function(x, shift_by = 1) {
   stopifnot(is.numeric(shift_by))
   stopifnot(is.numeric(x))

cran-comments.md

@@ -1,60 +1,56 @@
-## Resubmission after archiving
-
-This is the first submission of `ggRandomForests` since the package was
-archived from CRAN on **2025-07-01**. The archiving notice was *"issues
-were not corrected in time."* The underlying issue was an upstream API
-change: `randomForestSRC` removed its `var.select.rfsrc()` workflow that
-several `ggRandomForests` helpers depended on. v2.7.1 resolves this by:
-
-* Removing the `var.select`-based variable-selection paths.  Users are
-  now directed to use minimum depth (already supported via
-  `randomForestSRC::max.subtree`) or VIMP-based ranking (`gg_vimp`) until
-  a `varPro`-based replacement is fleshed out in a future release.
-* Updating all vignettes and examples to use the supported workflow.
-* Fixing a separate batch of latent bugs documented in `NEWS.md` for
-  v2.7.0 / v2.7.1 (S3 design overhaul, empty-figure bugs in survival
-  partial-dependence, duplicate VIMP legend, NEWS-source consolidation,
-  PDF-manual Unicode-minus issue, etc.).
+## v2.7.2 — Resubmission addressing CRAN reviewer feedback
+
+This is a resubmission of v2.7.1 addressing every item raised by
+Benjamin Altmann's review of 2026-04-27. All four requested changes have
+been applied:
+
+1. **References in DESCRIPTION.** Added Breiman (2001)
+   `<doi:10.1023/A:1010933404324>` and Ishwaran et al. (2008)
+   `<doi:10.1214/08-AOAS169>` to the Description field.
+2. **`:::` in documentation.** `man/shift.Rd` had an example using
+   `ggRandomForests:::shift(...)`. `shift()` is an internal utility, so
+   the help page has been removed entirely (the function is now
+   `@noRd`); the `:::` reference is gone.
+3. **`cat()` in `R/surv_partial.rfsrc.R`.** Replaced the user-visible
+   `cat("partial plot for: ", xvar, "\n")` with `message(...)` so the
+   output can be suppressed via `suppressMessages()` and plays nicely
+   inside notebooks / Shiny / quarto.
+4. **`par()` reset in `man/surv_partial.rfsrc.Rd` example.** Wrapped
+   the `par(mfrow = c(2, 2))` block with
+   `oldpar <- par(no.readonly = TRUE); on.exit(par(oldpar))` so the
+   user's graphical parameters are restored on exit.
+
+## Background — first resubmission since archiving
+
+`ggRandomForests` was archived from CRAN on **2025-07-01** with the
+notice *"issues were not corrected in time."* The root cause was an
+upstream API change: `randomForestSRC` removed its
+`var.select.rfsrc()` workflow that several `ggRandomForests` helpers
+depended on. The fix in v2.7.x is to drop the `var.select`-based
+variable-selection paths and direct users to minimum depth
+(`randomForestSRC::max.subtree`) or VIMP-based ranking (`gg_vimp`)
+until a `varPro`-based replacement is fleshed out in a future release.
 
 ## Test environments
 
 * **Local:** R 4.5.3 on macOS Tahoe 26.4.1 (aarch64-apple-darwin20).
   `R CMD check --as-cran` returns 0 errors, 0 warnings, 2 informational
   NOTEs (CRAN incoming feasibility + local NTP timestamp; neither
   actionable on CRAN's own machines).
-* **win-builder R-release** (R 4.6.0, x86_64-w64-mingw32): 2 NOTEs
-  (see disposition below).
-* **win-builder R-devel:** submitted; results pending at the time of
-  this submission.
 * **GitHub Actions matrix:** ubuntu-latest (R-devel / R-release /
   R-oldrel-1), windows-latest (R-release), macos-latest (R-release) —
   all green on the head commit.
+* **Win-builder R-release / R-oldrel** (v2.7.1 tarball, prior to the
+  reviewer feedback): both clean — Status 1 NOTE, the new-submission
+  NOTE only.
 * **Reverse-dependency check:** `tools::package_dependencies(reverse =
-  TRUE)` returns 0; no downstream CRAN packages depend on
-  `ggRandomForests`.
+  TRUE)` returns 0; CRAN's auto-pretest also reported "No strong
+  reverse dependencies to be checked".
 * **URLs:** `urlchecker::url_check()` clean.
 
 ## NOTE disposition
 
-### NOTE 1 — "New submission / Package was archived on CRAN"
+### NOTE — "New submission / Package was archived on CRAN"
 
-Expected for a resubmission. The archiving root cause and its
-remediation in v2.7.1 are documented above.
-
-### NOTE 2 — Examples > 10s on win-builder R-release
-
-Win-builder R-release flagged two examples that exceeded 10 s
-cumulative CPU + elapsed (`gg_variable` 18.4 s; `gg_rfsrc.rfsrc`
-10.4 s). Both have been trimmed in this submission: the headline
-runnable example is now a small `ntree = 50` classification fit on
-`iris`, and the larger regression / survival demonstrations are
-guarded with `\donttest{}` so they only run under `--run-donttest`.
-
-## Other notes
-
-* `cran-comments.md` reflects the disposition documented above; no
-  silent NOTEs remain.
-* `NEWS.md` carries entries for both v2.7.0 (the API-change response)
-  and v2.7.1 (bug-fix follow-ups). The v2.7.0 entry is included
-  because that release was tagged in the source repository but never
-  reached CRAN before archiving.
+Expected for a resubmission after archiving. The root cause and
+remediation in v2.7.x are documented above.

man/shift.Rd

@@ -187,8 +187,9 @@ local({
     expect_false(identical(yhat_mort, yhat_surv))
   })
 
-  test_that("surv_partial.rfsrc verbose: prints variable name during computation", {
-    expect_output(
+  test_that("surv_partial.rfsrc verbose: emits a message with the variable name", {
+    # v2.7.2: cat() -> message() so output is suppressible per CRAN cookbook.
+    expect_message(
       suppressWarnings(
         surv_partial.rfsrc(v.obj, var_list = "age", partial.type = "mort")
       ),