WIP Benchmark script

data-raw/benchmark_update_snapshot.R

@@ -0,0 +1,55 @@
+devtools::load_all()
+
+# Our benchmark data is the iris data set but repeated to increase the data size
+data_generator <- function(repeats) {
+  purrr::map(
+    seq(repeats),
+    \(it) dplyr::mutate(iris, r = dplyr::row_number() + (it - 1) * nrow(iris))
+  ) |>
+    purrr::reduce(rbind)
+}
+
+data_1 <- data_generator(1)
+data_2 <- data_generator(2) |>
+  dplyr::mutate(
+    Sepal.Length = dplyr::if_else(Sepal.Length > median(Sepal.Length), Sepal.Length, Sepal.Length / 2)
+  )
+data_3 <- data_generator(3) |>
+  dplyr::mutate(
+    Sepal.Length = dplyr::if_else(Sepal.Length > median(Sepal.Length), Sepal.Length, Sepal.Length / 2),
+    Sepal.Width = dplyr::if_else(Sepal.Width > median(Sepal.Width), Sepal.Width, Sepal.Width / 2)
+  )
+
+conns <- get_test_conns()
+conn <- conns[[1]]
+
+# copy data to the conns
+data_on_conn <- list(
+  dplyr::copy_to(conn, data_1, name = id("test.SCDB_data_1", conn), overwrite = TRUE, temporary = FALSE),
+  dplyr::copy_to(conn, data_2, name = id("test.SCDB_data_2", conn), overwrite = TRUE, temporary = FALSE),
+  dplyr::copy_to(conn, data_3, name = id("test.SCDB_data_3", conn), overwrite = TRUE, temporary = FALSE)
+)
+
+# Define the data to loop over for benchmark
+ts <- list("2021-01-01", "2021-01-02", "2021-01-03")
+
+# Define the SCDB update functions
+scdb_update_step <- function(conn, data, ts) {
+  update_snapshot(data, conn, "SCDB_benchmark", timestamp = ts,
+                  logger = Logger$new(output_to_console = FALSE, warn = FALSE))
+}
+
+scdb_updates <- function(conn, data_on_conn) {
+  purrr::walk2(data_on_conn, ts, \(data, ts) scdb_update_step(conn, data, ts))
+  DBI::dbRemoveTable(conn, name = "SCDB_benchmark")
+}
+
+# Construct the list of benchmarks
+benchmark_exprs <- alist()
+benchmark_on_conn <- local(bquote(scdb_updates(conn, data_on_conn)))
+benchmark_exprs <- append(benchmark_exprs, benchmark_on_conn)
+names(benchmark_exprs) <- names(conns)[[1]]
+
+res <- microbenchmark::microbenchmark(list = benchmark_exprs, times = 10)
+
+print(res)