perf(cran): de-link timing-out data URLs + skip Suggests-backed Windows tests (v0.9.155)

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: soilKey
 Type: Package
 Title: Automated Soil Profile Classification per WRB 2022, 'SiBCS' 5 and USDA Soil Taxonomy 13
-Version: 0.9.154
+Version: 0.9.155
 Date: 2026-06-21
 Authors@R:
     person("Hugo", "Rodrigues",

NEWS.md

@@ -1,3 +1,24 @@
+# soilKey 0.9.155 (2026-06-21)
+
+Final check-time trim (the 0.9.154 pre-test was OK on Debian and on Windows but
+the Windows overall check time was 12 min, just over the 10 min target). No
+classification change; 44 canonical fixtures byte-identical.
+
+\itemize{
+  \item \strong{External data-server URLs are no longer checked links.} The Rd
+        `\\url{}` entries (USDA, ISRIC, MapBiomas, ESDAC, Embrapa, FEBR, ...) and
+        the README/vignette links to those servers are now plain code spans.
+        One of them (MapBiomas) timed out for the check host on every run and
+        alone added ~60 s to the feasibility step; the addresses are still
+        visible, just not pinged.
+  \item \strong{The Suggests-backed integration tests} (aqp interop, the QGIS
+        export, the ESDB raster reader, the Munsell-prediction suite) now
+        `skip_on_cran()`. They run on Windows where those Suggests are present,
+        which is why they did not appear in local timings; they run in full on
+        CI. The classification keys, diagnostic predicates and 44 fixtures still
+        run on CRAN.
+}
+
 # soilKey 0.9.154 (2026-06-21)
 
 The "**lean check**" release: a real reduction of the CRAN check footprint
@@ -6794,7 +6815,7 @@ mapping layer:
 - **`classify_via_smartsolos_api(pedon, api_key, endpoint,
   drenagem, reference_sibcs, base_url, timeout_seconds, post_fn,
   verbose)`** -- POSTs a soilKey \code{PedonRecord} to
-  \code{https://api.cnptia.embrapa.br/smartsolos/expert/v1/classification}
+  \code{`https://api.cnptia.embrapa.br/smartsolos/expert/v1/classification`}
   (or \code{/verification}) and returns a
   \code{ClassificationResult} with the Embrapa-hosted Ordem /
   Subordem / Grande Grupo / Subgrupo. Bearer token comes from

R/afsp.R

@@ -169,7 +169,7 @@ wrb06_code_to_rsg <- function(code) {
 #'
 #' Reads the AfSP DBase tables shipped inside \code{AF-AfSP1.2.zip}
 #' (downloadable from
-#' \url{https://files.isric.org/public/afsp/AF-AfSP1.2.zip}) and
+#' \code{https://files.isric.org/public/afsp/AF-AfSP1.2.zip}) and
 #' converts each profile + its horizons to a soilKey
 #' \code{\link{PedonRecord}}. Filters to profiles with a populated
 #' WRB 2006 RSG code (i.e.\ classifiable; AfSP has ~7000 of these of
@@ -193,7 +193,7 @@ wrb06_code_to_rsg <- function(code) {
 #' (2014). Africa Soil Profiles Database, Version 1.2. ISRIC Report
 #' 2014/01. ISRIC -- World Soil Information, Wageningen.
 #' Project page:
-#' \url{https://isric.org/projects/africa-soil-profiles-database-afsp}.
+#' \code{https://isric.org/projects/africa-soil-profiles-database-afsp}.
 #'
 #' @export
 load_afsp_pedons <- function(afsp_dir,

R/benchmark-loaders.R

@@ -176,7 +176,7 @@ load_lucas_pedons <- function(lucas_csv, head = NULL, verbose = TRUE) {
 #' The dadosolos / BDsolos archive ships with ~5k profiles in PT-BR
 #' with full SiBCS classification, lab data, and horizon morphology --
 #' the primary validation set for Brazilian-context use. Available
-#' from \url{https://www.bdsolos.cnptia.embrapa.br/}.
+#' from \code{https://www.bdsolos.cnptia.embrapa.br/}.
 #'
 #' @param csv_path Path to the BDsolos CSV (long format: one row per
 #'        horizon, with a profile-id key and per-profile classification).

R/benchmark-lucas-2018.R

@@ -163,7 +163,7 @@
 #' Reads the canonical European Soil Data Centre (ESDAC) release of
 #' LUCAS Soil 2018 Topsoil chemistry as published in the JRC report
 #' (ESDAC dataset
-#' \url{https://esdac.jrc.ec.europa.eu/content/lucas-2018-topsoil-data}).
+#' \code{https://esdac.jrc.ec.europa.eu/content/lucas-2018-topsoil-data}).
 #' The release ships ~18,984 European topsoil samples at 0-20 cm with
 #' pH (H2O and CaCl2), EC, OC, CaCO3, P, N, K and oxalate-extractable
 #' Al / Fe; a separate \code{BulkDensity_2018_final-2.csv} carries

R/classify-smartsolos.R

@@ -295,7 +295,7 @@
 #' \code{L0..L4}).
 #'
 #' Authentication: register a free AgroAPI account at
-#' \url{https://www.agroapi.cnptia.embrapa.br/portal/}, subscribe to
+#' \code{https://www.agroapi.cnptia.embrapa.br/portal/}, subscribe to
 #' the SmartSolosExpert API and generate an access token. Pass it via
 #' the \code{AGROAPI_TOKEN} environment variable or the
 #' \code{api_key} argument.

R/data-ossl-demo.R

@@ -46,7 +46,7 @@
 #' Synthetic; built by \code{data-raw/build_ossl_demo.R} with seed
 #' 20260430. The OSSL property ranges that drove the simulation
 #' come from Sanderman, J. \emph{et al.} (2024), \emph{Open Soil
-#' Spectral Library}, \url{https://soilspectroscopy.org/}.
+#' Spectral Library}, \code{https://soilspectroscopy.org/}.
 #'
 #' @examples
 #' data(ossl_demo_sa)

R/febr.R

@@ -159,7 +159,7 @@
 #'
 #' The FEBR \code{camada} (layer) table uses standardised variable
 #' codes documented in the FEBR data dictionary (see
-#' \url{https://www.pedometria.org/febr/} for the project home;
+#' \code{https://www.pedometria.org/febr/} for the project home;
 #' the dictionary path moved during 2024 -- the codes themselves
 #' are stable). This internal table records the regex patterns that
 #' map the most useful FEBR codes onto the soilKey horizon schema.

R/soil-classes-at-location.R

@@ -40,7 +40,7 @@
 #' @section Data source:
 #' For real use, point \code{source_url} at a regional SoilGrids
 #' "MostProbable WRB" GeoTIFF / COG (one of the cuts at
-#' \url{https://files.isric.org/soilgrids/latest/data/wrb/}). For
+#' \code{https://files.isric.org/soilgrids/latest/data/wrb/}). For
 #' tests, \code{options(soilKey.test_raster = "/tmp/syn.tif")} is
 #' honoured. When no source is given, the function emits a
 #' \code{cli_alert_warning()} and returns an empty result -- it does

R/spatial-embrapa.R

@@ -34,7 +34,7 @@
 #' @param raster_path Required. Path to a local categorical raster
 #'        (GeoTIFF) of Embrapa SiBCS classes. There is no built-in
 #'        file in v0.5 -- download the polygon map from
-#'        \url{https://www.embrapa.br/solos/sibcs} and rasterise it.
+#'        \code{https://www.embrapa.br/solos/sibcs} and rasterise it.
 #' @param buffer_m Buffer radius in metres (default 3750, i.e.
 #'        ~15-cell neighbourhood at 250 m resolution).
 #' @param lut Optional named character vector mapping raster integer

R/spatial-lookups.R

@@ -27,7 +27,7 @@
 #'
 #' MapBiomas Solos (Project MapBiomas, Brazil) distributes a national
 #' raster of SiBCS classes at 30 m, downloadable from
-#' \url{https://mapbiomas.org/en/produtos}. This helper mirrors the
+#' \code{https://mapbiomas.org/en/produtos}. This helper mirrors the
 #' shape of \code{\link{lookup_esdb}} but is local-file only: pass
 #' the path of the unpacked GeoTIFF and the function reprojects the
 #' user's WGS84 lat/lon to the raster's native CRS, extracts the
@@ -111,7 +111,7 @@ lookup_mapbiomas_solos <- function(coords, raster_path, legend = NULL) {
 #'
 #' Reads ISRIC SoilGrids 250m (Hengl et al. 2017, 2021) directly
 #' from the ISRIC Cloud-Optimized GeoTIFF (COG) endpoint at
-#' \url{https://files.isric.org/soilgrids/latest/data/} -- no
+#' \code{https://files.isric.org/soilgrids/latest/data/} -- no
 #' download required, only the pixel under each query coordinate is
 #' transferred over HTTPS.
 #'

R/spectra-fetch.R

@@ -72,7 +72,7 @@
 #' Sanderman, J., Savage, K., Dangal, S.R.S., Duran, G., Rivard, C.,
 #' Cardona, M.T., Sandzhieva, A., Aramian, A. & Safanelli, J.L. (2024).
 #' Soil Spectroscopy for Global Good -- the Open Soil Spectral Library
-#' (OSSL). \url{https://soilspectroscopy.org/}.
+#' (OSSL). \code{https://soilspectroscopy.org/}.
 #'
 #' @export
 download_ossl_subset <- function(region      = c("global", "south_america",

R/wosis-cache.R

@@ -184,7 +184,7 @@ load_wosis_stratified_sample <- function() {
 #' @section Reference:
 #' Beaudette, D., Skovlin, J., Roecker, S., Brown, A. (2024). aqp:
 #' Algorithms for Quantitative Pedology. R package version 2.x.
-#' \url{https://github.com/ncss-tech/aqp}.
+#' \code{https://github.com/ncss-tech/aqp}.
 #'
 #' @examples
 #' \dontrun{
@@ -281,7 +281,7 @@ load_kssl_sample <- function() {
 #' @section Reference:
 #' Beaudette, D., Skovlin, J., Roecker, S., Brown, A. (2024). aqp:
 #' Algorithms for Quantitative Pedology. R package version 2.x.
-#' \url{https://github.com/ncss-tech/aqp}.
+#' \code{https://github.com/ncss-tech/aqp}.
 #'
 #' @examples
 #' \dontrun{

README.md

@@ -519,12 +519,12 @@ When you use **`benchmark_redape()`** or **`load_redape_pedons()`**:
 
 ## ✦ References (canonical books + datasets)
 
-* **WRB 2022** — IUSS Working Group WRB (2022). *World Reference Base for Soil Resources, 4th edition.* International Union of Soil Sciences, Vienna, Austria. [FAO OpenKnowledge PDF](https://openknowledge.fao.org/server/api/core/bitstreams/bcdecec7-f45f-4dc5-beb1-97022d29fab4/content)
+* **WRB 2022** — IUSS Working Group WRB (2022). *World Reference Base for Soil Resources, 4th edition.* International Union of Soil Sciences, Vienna, Austria. FAO OpenKnowledge PDF (`https://openknowledge.fao.org/server/api/core/bitstreams/bcdecec7-f45f-4dc5-beb1-97022d29fab4/content`)
 * **SiBCS 5** — Santos, H. G. *et al.* (2018). *Sistema Brasileiro de Classificação de Solos*, 5th revised and extended edition. Embrapa, Brasília.
 * **USDA Soil Taxonomy 13** — Soil Survey Staff (2022). *Keys to Soil Taxonomy*, 13th edition. USDA-NRCS, Washington, DC.
 * **OSSL** — Sanderman, J., Savage, K., & Dangal, S. R. S. (2020). *Mid-infrared spectroscopy for prediction of soil health indicators in the United States.* Soil Science Society of America Journal, 84(1), 251–261.
 * **WoSIS** — Batjes, N. H., Ribeiro, E., & van Oostrum, A. (2020). *Standardised soil profile data to support global mapping and modelling (WoSIS snapshot 2019).* Earth System Science Data, 12, 299–320. <https://doi.org/10.5194/essd-12-299-2020>
-* **AfSP** — Leenaars, J. G. B., van Oostrum, A. J. M., & Ruiperez Gonzalez, M. (2014). *Africa Soil Profiles Database, Version 1.2.* ISRIC Report 2014/01. ISRIC — World Soil Information, Wageningen. [Project page](https://isric.org/projects/africa-soil-profiles-database-afsp). The bundled `afsp_sample.rds` is a 120-pedon stratified slice; `load_afsp_pedons()` parses the full upstream archive when available.
+* **AfSP** — Leenaars, J. G. B., van Oostrum, A. J. M., & Ruiperez Gonzalez, M. (2014). *Africa Soil Profiles Database, Version 1.2.* ISRIC Report 2014/01. ISRIC — World Soil Information, Wageningen. Project page (`https://isric.org/projects/africa-soil-profiles-database-afsp`). The bundled `afsp_sample.rds` is a 120-pedon stratified slice; `load_afsp_pedons()` parses the full upstream archive when available.
    *(Note: soilKey does not use the separate AfSIS — Africa Soil Information Service — soil property maps; only the ISRIC AfSP profile database.)*
 * **LUCAS 2018 — data report (this is what `benchmark_lucas_2018()` consumes)** — Fernandez-Ugalde, O., Scarpa, S., Orgiazzi, A., Panagos, P., Van Liedekerke, M., Marechal, A., & Jones, A. (2022). *LUCAS 2018 SOIL Component: sampling intensity, harmonisation and procedures for the collection of soil samples.* JRC Technical Report 130218, European Commission, Joint Research Centre, Ispra. <https://doi.org/10.2760/215013>
 * **LUCAS 2018 — review** — Orgiazzi, A., Ballabio, C., Panagos, P., Jones, A., & Fernández-Ugalde, O. (2018). *LUCAS Soil, the largest expandable soil dataset for Europe: a review.* European Journal of Soil Science, 69(1), 140–153. <https://doi.org/10.1111/ejss.12499>

cran-comments.md

@@ -1,46 +1,47 @@
-# cran-comments.md -- soilKey 0.9.154
+# cran-comments.md -- soilKey 0.9.155
 
 ## Resubmission
 
-This is a resubmission. The 0.9.154 incoming pre-test was **OK on both Windows
-and Debian**, but was held for `Overall checktime 16 min > 10 min`. This version
-is a genuine reduction of the check footprint (not a wholesale disabling of
-tests), and addresses the two NOTEs:
-
-* **Manual size (the main lever):** the ~600 internal rule-engine predicates
-  were already marked `@keywords internal`; they are now *truly* internal --
-  removed from `NAMESPACE` and the reference manual (the rule engine resolves
-  them from the namespace; tests reach them via `test_check()`). The documented
-  surface drops from **928 to 319 topics** (man pages 1129 -> 335), which is
-  what cuts the HTML/PDF manual build. The public API (`classify_*`,
-  `PedonRecord`, `report*`, `coverage_report`, the benchmark and gap-fill entry
-  points, the documented diagnostics) is unchanged and classification is
-  **byte-identical** (44 canonical fixtures).
-
-* **URL NOTE:** the USDA-NRCS NCSS data-mart address (which timed out for the
-  check host, ~100 s of the feasibility step) is no longer a checked link.
-
-* **Performance test (per the reviewer's question):** the wall-clock
-  "< 5 s/pedon" assertion -- the source of the released 0.9.96 ATLAS WARNING --
-  is **removed, not disabled**. The test now runs on CRAN and verifies the
-  benchmark returns well-formed, non-negative timings; an absolute wall-clock
-  threshold is not portable across CRAN's CPU/BLAS variants, so the
-  speed-regression guard now lives in CI.
-
-* Long-running benchmark, simulation, spectral, vision-language and spatial
-  tests (which need optional Suggests or external data) are conditioned off
-  CRAN with `skip_on_cran()` -- because they are long-running per "Writing R
-  Extensions", not because they fail. The classification keys, the diagnostic
-  predicates and the 44 fixtures still run on CRAN.
+This is a resubmission. The 0.9.154 pre-test was **OK on Debian and OK on
+Windows** but the Windows overall check time was 12 min (target 10 min). The
+Windows log showed where the remaining time went, and 0.9.155 trims it:
+
+* **A data-server URL was timing out.** `checking CRAN incoming feasibility`
+  took 142 s on Windows because the MapBiomas link in
+  `lookup_mapbiomas_solos.Rd` timed out (60 s, server unreachable from the
+  check host). All `\url{}` entries to external soil-data servers (USDA, ISRIC,
+  MapBiomas, ESDAC, Embrapa, FEBR, ...) -- and the matching README/vignette/NEWS
+  links -- are now plain code spans: the addresses are still shown, just not
+  pinged. Feasibility drops to ~30 s.
+
+* **Suggests-backed integration tests ran only on the Windows host.** The aqp
+  interop, QGIS export, ESDB raster and Munsell-prediction suites need optional
+  Suggests that are present on win-builder but not on my machine, so they
+  inflated the Windows test phase invisibly. They are now `skip_on_cran()` and
+  run in full on CI.
+
+The 0.9.154 reductions are retained:
+
+* The ~600 internal rule-engine predicates, already marked `@keywords internal`,
+  are no longer exported and no longer documented (928 -> 319 topics; man pages
+  1129 -> 335), which cut the HTML/PDF manual build. The public API is unchanged
+  and classification is **byte-identical** (44 canonical fixtures).
+* The performance test's wall-clock "< 5 s/pedon" assertion (the source of the
+  released 0.9.96 ATLAS WARNING) was **removed, not disabled** -- the test runs
+  on CRAN and checks the timings are well-formed; the speed guard is in CI.
+* Long-running benchmark/simulation/spectral/vision-language/spatial tests are
+  `skip_on_cran()` (run on CI) -- long-running per "Writing R Extensions", not
+  failing. The classification keys, diagnostic predicates and 44 fixtures still
+  run on CRAN.
 
 A full local `R CMD check --as-cran` (with vignettes and manual) is
-0 errors / 0 warnings / 1 NOTE; the testthat phase is ~75 s and the manual
-build is a fraction of its former size.
+0 errors / 0 warnings / 1 NOTE; feasibility ~30 s, testthat ~60 s, and the
+manual build is a fraction of its former size.
 
 ## Submission summary
 
 This is a maintenance update to soilKey, following the v0.9.96 submission on
-2026-05-19 (currently on CRAN). The v0.9.97 -> v0.9.154 series is
+2026-05-19 (currently on CRAN). The v0.9.97 -> v0.9.155 series is
 backward-compatible and tracked in NEWS per release. Every new feature is
 additive and **default off**, so a v0.9.96 call returns byte-identical output;
 44 canonical classification fixtures are regression-locked across the series.
@@ -101,7 +102,7 @@ new argument has a safe default.
 - macos-latest / R-release; windows-latest / R-release (GitHub Actions).
 - pkgdown (`pkgdown::check_pkgdown()`) and test-coverage (separate workflows).
 
-## R CMD check --as-cran results (v0.9.154)
+## R CMD check --as-cran results (v0.9.155)
 
 A full local build (with vignettes and manual) and `R CMD check --as-cran`:
 
@@ -112,9 +113,9 @@ A full local build (with vignettes and manual) and `R CMD check --as-cran`:
   local-only artefact -- the macOS-bundled `tidy` is older than CRAN's -- and
   does not appear on the CRAN check servers. "CRAN incoming feasibility",
   "checking examples (+ --run-donttest)", and "re-building of vignette outputs"
-  (all 13 vignettes) are **OK**. The full test suite runs on CI; on CRAN the heavy tests skip, so the testthat phase is ~75 s.
+  (all 13 vignettes) are **OK**. The full test suite runs on CI; on CRAN the heavy tests skip, so the testthat phase is ~60 s.
 
-All tests pass under `R CMD check` (`testthat.R`, ~75 s on CRAN with the
+All tests pass under `R CMD check` (`testthat.R`, ~60 s on CRAN with the
 long-running suites skipped; full suite on CI). Tests that need optional
 Suggests (magick, jsonvalidate, pdftools, ellmer, terra, sf, aqp, prospectr,
 resemble, ...) use `skip_if_not_installed()`.