data frame equality test with ignore_row_order=TRUE detects differenc…

…e in number of rows. closes tidyverse#1065

NEWS.md

@@ -12,7 +12,9 @@
 * Fixed issue about complex variables used in `summarise` (#933). 
 
 * Set operations give more useful error message on incompatible data frames (#903). 
-
+
+* `all.equal` gives the correct result when `ignore_row_order` is `TRUE` (#1065). 
+
 # dplyr 0.4.1
 
 * Don't assume that RPostgreSQL is available.

src/dplyr.cpp

@@ -1413,9 +1413,12 @@ dplyr::BoolResult equal_data_frame(DataFrame x, DataFrame y, bool ignore_col_ord
 
     // train the map in both x and y
     int nrows_x = x.nrows() ;
-    for( int i=0; i<nrows_x; i++) map[i].push_back(i) ;
-
     int nrows_y = y.nrows() ;
+
+    if( nrows_x != nrows_y )
+        return no_because( "Different number of rows" ) ;
+
+    for( int i=0; i<nrows_x; i++) map[i].push_back(i) ;
     for( int i=0; i<nrows_y; i++) map[-i-1].push_back(-i-1) ;
 
     RowTrack track_x( "Rows in x but not y: " ) ;
@@ -1457,8 +1460,6 @@ dplyr::BoolResult equal_data_frame(DataFrame x, DataFrame y, bool ignore_col_ord
     if(ok && ignore_row_order) return yes();
 
     if( !ignore_row_order ){
-        if( nrows_x != nrows_y )
-            return no_because( "Different number of rows" ) ;
         for( int i=0; i<nrows_x; i++){
             if( !visitors.equal( i, -i-1) ){
                     return no_because( "Same row values, but different order" ) ;

tests/testthat/test-equality.r

@@ -67,3 +67,9 @@ test_that("all.equal.data.frame handles data.frames with NULL names", {
   names(x) <- NULL
   expect_true(all.equal(x,x))
 })
+
+test_that( "data frame equality test with ignore_row_order=TRUE detects difference in number of rows. #1065", {
+  DF1 <- data_frame(a = 1:4, b = letters[1:4])
+  DF2 <- data_frame(a = c(1:4,4L), b = letters[c(1:4,4L)])
+  expect_false( isTRUE(all.equal(DF1, DF2, ignore_row_order=TRUE)))
+})