Add ad-hoc tests

.Rbuildignore

@@ -19,3 +19,4 @@
 ^\.binder$
 ^\.vscode$
 ^CRAN-SUBMISSION$
+^ad-hoc-tests$

R/od-funs.R

@@ -77,8 +77,14 @@ od_to_sfc = function(x,
 
 #' Create matrices representing origin-destination coordinates
 #'
-#' This function takes a wide range of input data types (spatial lines, points or text strings)
-#' and returns a data frame of coordinates representing origin (ox, oy) and destination (dx, dy) points.
+#' This function takes an 'od data frame' with the first
+#' two columns matching IDs of spatial objects, and
+#' matches them with objects representing origins and destinations
+#' in wide range of input data types (spatial lines, points or text strings).
+#' It returns a data frame of coordinates representing movement between all origin (ox, oy) and destination (dx, dy) points.
+#' 
+#' See [points_to_od()] for a function that creates
+#' an 'od data frame' from a set (or two sets) of points.
 #' @param p Points representing origins and destinations
 #' @param pd Points representing destinations, if different from origin points
 #' @param sfnames Should output column names be compatible with the sf package?

ad-hoc-tests/.gitignore

@@ -0,0 +1 @@
+/.quarto/

ad-hoc-tests/test-max-dist-speedup.qmd

@@ -0,0 +1,69 @@
+---
+format: gfm
+---
+
+This document tests the new `max-dist` functionality in PR [#48](https://github.com/ITSLeeds/od/pull/48).
+
+Let's start the test documented in the PR with the installed version of the package.
+
+```{r}
+remotes::install_cran("od")
+library(sf)
+```
+
+# Test 1: 1000 points
+
+
+```{r}
+p = pct::get_centroids_ew()
+p = p[1:1000,]
+
+system.time(r1 <- od::points_to_od(p))
+head(r1)
+nrow(r1)
+```
+
+Now let's test the new `max-dist` functionality.
+
+
+```{r}
+if (!file.exists("DESCRIPTION")) {
+  setwd("..")
+}
+devtools::load_all()
+system.time(r2 <- points_to_od(p))
+head(r2)
+nrow(r2)
+```
+
+
+```{r}
+system.time(r3 <- points_to_od(p, max_dist = 1000))
+head(r3)
+nrow(r3)
+```
+
+The benchmark shows that the new `max-dist` functionality is faster than the original implementation for large datasets.
+
+Let's compare the results.
+
+
+```{r}
+waldo::compare(head(r1), head(r2))
+r2_sorted = r2 |>
+    dplyr::arrange(desc(O), desc(D)) 
+r1_sorted = r1 |>
+    dplyr::arrange(desc(O), desc(D))
+waldo::compare(head(r1_sorted), head(r2_sorted))
+```
+
+Let's plot the results for the max-dist = 1000 case.
+
+
+```{r}
+r3_sf = od::od_to_sf(r3, p)
+plot(sf::st_geometry(p), col = "red")
+plot(sf::st_geometry(r3_sf), add = TRUE)
+```
+
+# Test 2: od_coordinates