testmatrix.c: Don’t use g_assert_cmphex() with floatstestmatrix-float

The build was giving this warning:
../tests/testmatrix.c:197:27: warning: implicit conversion from 'double' to 'guint64' (aka 'unsigned long long') changes value from 0.2 to 0 [-Wliteral-conversion]
  g_assert_cmphex (r, ==, 0.2);
  ~~~~~~~~~~~~~~~~~~~~~~~~^~~~

So the test was effectively comparing 0 to 0.

I switched to g_assert_cmpfloat(), but now the tests are failing as the
values don’t mach. I don’t know which is wrong, the expected value or
the code being tested.

1 files changed, 4 insertions, 4 deletions

diff --git a/tests/testmatrix.c b/tests/testmatrix.c
index 1ba79839..3bbca282 100644
--- a/tests/testmatrix.c
+++ b/tests/testmatrix.c
@@ -194,20 +194,20 @@ test_matrix_slant_ratio (void)
   double r;
 
   r = pango_matrix_get_slant_ratio (&m);
-  g_assert_cmphex (r, ==, 0.2);
+  g_assert_cmpfloat (r, ==, 0.2);
 
   pango_matrix_rotate (&m, 45);
 
   r = pango_matrix_get_slant_ratio (&m);
-  g_assert_cmphex (r, ==, 0.2);
+  g_assert_cmpfloat (r, ==, 0.2);
 
   pango_matrix_scale (&m, 2, 3);
 
   r = pango_matrix_get_slant_ratio (&m);
-  g_assert_cmphex (r, ==, 0.2);
+  g_assert_cmpfloat (r, ==, 0.2);
 
   r = pango_matrix_get_slant_ratio (&m2);
-  g_assert_cmphex (r, ==, 0.4);
+  g_assert_cmpfloat (r, ==, 0.4);
 }
 
 int