1 files changed, 49 insertions, 0 deletions
diff --git a/internal/dup/dup_test.go b/internal/dup/dup_test.go
index 8984336..a616118 100644
--- a/internal/dup/dup_test.go
+++ b/internal/dup/dup_test.go
@@ -577,3 +577,52 @@ func TestThreeCopiesWithOverlappingExtraDirKeepOne(t *testing.T) {
 		}
 	}
 }
+
+// TestLookupWalksAClassOnce: every member of a content class elects the
+// same original, so the class is worth computing once. Walking the size
+// group again for every member makes a directory of N copies cost N^2
+// comparisons - 8000 identical files took 35 s of it - and the walk takes
+// the index's lock at every step, so more workers made it slower rather
+// than faster. This counts the walks: one per class, not one per file.
+func TestLookupWalksAClassOnce(t *testing.T) {
+	d := t.TempDir()
+	const n = 12
+	var files []scan.File
+	for i := 0; i < n; i++ {
+		// Two classes of one size, so the size group cannot be the thing
+		// being memoised.
+		body := "the same bytes for all of these"
+		if i%2 == 1 {
+			body = "different bytes, identical size"
+		}
+		files = append(files, put(t, d, fmt.Sprintf("f%02d.bin", i), []byte(body), i))
+	}
+	x, errs := NewIndex(files, nil)
+	if len(errs) > 0 {
+		t.Fatal(errs)
+	}
+	for _, f := range files {
+		lookup(t, x, f)
+	}
+	if x.walks > 2 {
+		t.Errorf("identicalTo ran %d times for 2 content classes over %d files; want one walk per class", x.walks, n)
+	}
+
+	// The answers must be what they were before memoising: one original
+	// per class, and that original is not itself a duplicate.
+	dups := 0
+	origs := map[string]int{}
+	for _, f := range files {
+		orig, isDup := lookup(t, x, f)
+		if isDup {
+			dups++
+		}
+		origs[orig]++
+	}
+	if dups != n-2 {
+		t.Errorf("%d duplicates over two classes of %d files, want %d", dups, n, n-2)
+	}
+	if len(origs) != 2 {
+		t.Errorf("the two classes elected %d originals, want 2", len(origs))
+	}
+}