[caffe] 01/01: patch: new patch fix-more-float-comparison-precision-issue2

[Zhou Mo]

This is an automated email from the git hooks/post-receive script.

cdluminate-guest pushed a commit to branch master
in repository caffe.

commit 2e72f42dd238ffb8b0381b1f70db09dac806844e
Author: Zhou Mo <cdluminate at gmail.com>
Date:   Mon Nov 28 04:54:40 2016 +0000

    patch: new patch fix-more-float-comparison-precision-issue2
---
 debian/changelog                                          |  7 +++++++
 debian/patches/fix-more-float-comparison-precision-issue2 | 15 +++++++++++++++
 debian/patches/series                                     |  1 +
 3 files changed, 23 insertions(+)

diff --git a/debian/changelog b/debian/changelog
index d4deba1..acde139 100644
--- a/debian/changelog
+++ b/debian/changelog
@@ -1,3 +1,10 @@
+caffe (1.0.0~rc3+20160930-ga7f950b-4) UNRELEASED; urgency=medium
+
+  * Add patch: fix-more-float-comparison-precision-issue2 to fix more float
+    point number comparision issue.
+
+ -- Zhou Mo <cdluminate at gmail.com>  Mon, 28 Nov 2016 04:53:40 +0000
+
 caffe (1.0.0~rc3+20160930-ga7f950b-3) experimental; urgency=medium
 
   * Refresh symbols control file for i386 and ppc64el.
diff --git a/debian/patches/fix-more-float-comparison-precision-issue2 b/debian/patches/fix-more-float-comparison-precision-issue2
new file mode 100644
index 0000000..537e586
--- /dev/null
+++ b/debian/patches/fix-more-float-comparison-precision-issue2
@@ -0,0 +1,15 @@
+Purpose: fix float point number comparison precision issue to avoid FTBFS
+Forward: not-yet
+diff --git a/src/caffe/test/test_deconvolution_layer.cpp b/src/caffe/test/test_deconvolution_layer.cpp
+index c4b09ad..858a72c 100644
+--- a/src/caffe/test/test_deconvolution_layer.cpp
++++ b/src/caffe/test/test_deconvolution_layer.cpp
+@@ -257,7 +257,7 @@ TYPED_TEST(DeconvolutionLayerTest, TestNDAgainst2D) {
+   }
+   ASSERT_EQ(result_nd.count(), result_2d.count());
+   for (int i = 0; i < result_2d.count(); ++i)  {
+-    EXPECT_EQ(result_2d.cpu_data()[i], result_nd.cpu_data()[i]);
++    EXPECT_FLOAT_EQ(result_2d.cpu_data()[i], result_nd.cpu_data()[i]);
+   }
+   ASSERT_EQ(backward_result_nd.count(), backward_result_2d.count());
+   for (int i = 0; i < backward_result_2d.count(); ++i) {
diff --git a/debian/patches/series b/debian/patches/series
index 4805bd0..a1ab709 100644
--- a/debian/patches/series
+++ b/debian/patches/series
@@ -3,3 +3,4 @@ cmake-using-gnuinstalldirs
 cmake-fix-python-module-installdir
 cmake-avoid-argument-missing
 fix-more-float-comparison-precision-issue
+fix-more-float-comparison-precision-issue2

-- 
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-science/packages/caffe.git