- add tests to exercise hatch fills for floating images. This code is

  mis-generated with the gcc 4.0 that ships with OS X 10.4

diff --git a/Changes b/Changes
index 1a69720af301988a9cb2e64b98ba6756d76bc514..d6480beaec6a6593059b982d60e4ad2db49f77dc 100644 (file)
--- a/Changes
+++ b/Changes
@@ -1079,7 +1079,7 @@ Revision history for Perl extension Imager.
 - the FT1.x driver now supports the align parameter correctly.
   Tests were added to each driver to check correct handling of the align
   parameter.
-- the Win32 font driver bounding_box() method now puts accuarate values
+- the Win32 font driver bounding_box() method now puts accurate values
   in the ascent and descent values, previously they were set to the 
   font ascent/descent rather than the values specific to the string.
 - supplying align=>0 to the win32 font driver string() function 
@@ -1102,6 +1102,8 @@ Revision history for Perl extension Imager.
   compiler, MinGW can be used to build Imager.
 - fixed an inverted condition checking the giflib version, and 
   replaced a manual test skip in t/t105gif.t
+- add tests to exercise hatch fills for floating images.  This code is
+  mis-generated with the gcc 4.0 that ships with OS X 10.4
 
 =================================================================
 

diff --git a/t/t20fill.t b/t/t20fill.t
index f63bd27845c18dfc3a47e5b0ed8825660ebab3a1..146b05bfbec8898789c4b8ac3b0771f4a63c2251 100644 (file)
--- a/t/t20fill.t
+++ b/t/t20fill.t
@@ -1,6 +1,7 @@
 #!perl -w
 use strict;
-use Test::More tests => 47;
+use lib 't';
+use Test::More tests => 52;
 
 use Imager ':handy';
 use Imager::Fill;
@@ -12,6 +13,7 @@ Imager::init_log("testout/t20fill.log", 1);
 my $blue = NC(0,0,255);
 my $red = NC(255, 0, 0);
 my $redf = Imager::Color::Float->new(1, 0, 0);
+my $bluef = Imager::Color::Float->new(0, 0, 1);
 my $rsolid = Imager::i_new_fill_solid($blue, 0);
 ok($rsolid, "building solid fill");
 my $raw1 = Imager::ImgRaw::new(100, 100, 3);
@@ -68,6 +70,30 @@ Imager::i_box_cfill($raw2, 0, 0, 99, 99, $rcustom);
 $diff = Imager::i_img_diff($raw1, $raw2);
 ok(!$diff, "custom hatch mismatch");
 
+{
+  # basic test of floating color hatch fills
+  # this will exercise the code that the gcc shipped with OS X 10.4
+  # forgets to generate
+  # the float version is called iff we're working with a non-8-bit image
+  # i_new_fill_hatchf() makes the same object as i_new_fill_hatch() but
+  # we test the other construction code path here
+  my $fraw1 = Imager::i_img_double_new(100, 100, 3);
+  my $fhatch1 = Imager::i_new_fill_hatchf($redf, $bluef, 0, 1, undef, 0, 0);
+  ok($fraw1, "making double image 1");
+  ok($fhatch1, "making float hatch 1");
+  Imager::i_box_cfill($fraw1, 0, 0, 99, 99, $fhatch1);
+  my $fraw2 = Imager::i_img_double_new(100, 100, 3);
+  my $fhatch2 = Imager::i_new_fill_hatchf($bluef, $redf, 0, 1, undef, 0, 2);
+  ok($fraw2, "making double image 2");
+  ok($fhatch2, "making float hatch 2");
+  Imager::i_box_cfill($fraw2, 0, 0, 99, 99, $fhatch2);
+
+  $diff = Imager::i_img_diff($fraw1, $fraw2);
+  ok(!$diff, "float custom hatch mismatch");
+  save($fraw1, "testout/t20hatchf1.ppm");
+  save($fraw2, "testout/t20hatchf2.ppm");
+}
+
 # test the oo interface
 my $im1 = Imager->new(xsize=>100, ysize=>100);
 my $im2 = Imager->new(xsize=>100, ysize=>100);