Relax CI table comparison for cross-platform numeric drift

Author: Ekin-Kahraman

dev/compare_tables.R

@@ -7,7 +7,8 @@ if (length(args) != 2) {
 
 reference_dir <- args[[1]]
 generated_dir <- args[[2]]
-comparison_digits <- 5L
+relative_tolerance <- 2e-5
+zero_tolerance <- 1e-12
 
 if (!dir.exists(reference_dir)) {
   stop("Reference directory not found: ", reference_dir)
@@ -51,19 +52,35 @@ format_value <- function(value) {
 }
 
 compare_numeric_column <- function(reference_col, generated_col, file_label, col_name) {
-  reference_cmp <- signif(reference_col, digits = comparison_digits)
-  generated_cmp <- signif(generated_col, digits = comparison_digits)
-  mismatch <- first_mismatch(reference_cmp, generated_cmp)
-  if (!is.na(mismatch)) {
+  same <- (reference_col == generated_col) | (is.na(reference_col) & is.na(generated_col))
+  same[is.na(same)] <- FALSE
+
+  non_missing <- !(is.na(reference_col) | is.na(generated_col))
+  needs_check <- non_missing & !same
+  if (any(needs_check)) {
+    reference_vals <- reference_col[needs_check]
+    generated_vals <- generated_col[needs_check]
+    scale <- pmax(abs(reference_vals), abs(generated_vals))
+    allowed_diff <- ifelse(scale == 0, zero_tolerance, relative_tolerance * scale)
+    same[needs_check] <- abs(reference_vals - generated_vals) <= allowed_diff
+  }
+
+  mismatch <- which(!same)
+  if (length(mismatch) > 0) {
+    mismatch <- mismatch[[1]]
+    diff <- abs(reference_col[[mismatch]] - generated_col[[mismatch]])
+    scale <- max(abs(reference_col[[mismatch]]), abs(generated_col[[mismatch]]))
+    allowed_diff <- if (scale == 0) zero_tolerance else relative_tolerance * scale
     stop(
       sprintf(
-        "%s: numeric mismatch in column '%s' at row %d (reference=%s, generated=%s; compared at %d significant digits)",
+        "%s: numeric mismatch in column '%s' at row %d (reference=%s, generated=%s; abs_diff=%s, allowed=%s)",
         file_label,
         col_name,
         mismatch,
         format_value(reference_col[[mismatch]]),
         format_value(generated_col[[mismatch]]),
-        comparison_digits
+        format_value(diff),
+        format_value(allowed_diff)
       )
     )
   }
@@ -132,7 +149,7 @@ for (filename in reference_files) {
 message(
   "Tracked tables match regenerated outputs across ",
   length(reference_files),
-  " files at ",
-  comparison_digits,
-  " significant digits."
+  " files within relative tolerance ",
+  format(relative_tolerance, scientific = TRUE),
+  "."
 )