[RFC PATCH v2 10/12] test: libipa: Add Vector class test
Laurent Pinchart
laurent.pinchart at ideasonboard.com
Mon Nov 18 17:32:31 CET 2024
Hi Milan,
On Mon, Nov 18, 2024 at 02:27:59PM +0100, Milan Zamazal wrote:
> Laurent Pinchart <laurent.pinchart at ideasonboard.com> writes:
>
> > Add a unit test to exercize the API of the ipa::Vector class.
> >
> > The test binary being called 'vector', implicit includes cause the
> > binary to be picked by '#include <vector>', causing builds to fail. Set
> > implicit_include_directories to false to avoid this, as done in commit
> > 6cd849125888 ("test: Don't add current build directory to include
> > path").
> >
> > Signed-off-by: Laurent Pinchart <laurent.pinchart at ideasonboard.com>
> > ---
> > test/ipa/libipa/meson.build | 2 +
> > test/ipa/libipa/vector.cpp | 100 ++++++++++++++++++++++++++++++++++++
> > 2 files changed, 102 insertions(+)
> > create mode 100644 test/ipa/libipa/vector.cpp
> >
> > diff --git a/test/ipa/libipa/meson.build b/test/ipa/libipa/meson.build
> > index 4d2427dbd4e7..f9b3c46d000f 100644
> > --- a/test/ipa/libipa/meson.build
> > +++ b/test/ipa/libipa/meson.build
> > @@ -2,11 +2,13 @@
> >
> > libipa_test = [
> > {'name': 'interpolator', 'sources': ['interpolator.cpp']},
> > + {'name': 'vector', 'sources': ['vector.cpp']},
> > ]
> >
> > foreach test : libipa_test
> > exe = executable(test['name'], test['sources'],
> > dependencies : [libcamera_private, libipa_dep],
> > + implicit_include_directories : false,
> > link_with : [test_libraries],
> > include_directories : [test_includes_internal,
> > '../../../src/ipa/libipa/'])
> > diff --git a/test/ipa/libipa/vector.cpp b/test/ipa/libipa/vector.cpp
> > new file mode 100644
> > index 000000000000..5279bc8216b3
> > --- /dev/null
> > +++ b/test/ipa/libipa/vector.cpp
> > @@ -0,0 +1,100 @@
> > +/* SPDX-License-Identifier: GPL-2.0-or-later */
> > +/*
> > + * Copyright (C) 2024, Ideas on Board Oy
> > + *
> > + * Vector tests
> > + */
> > +
> > +#include "../src/ipa/libipa/vector.h"
> > +
> > +#include <cmath>
> > +#include <iostream>
> > +
> > +#include "test.h"
> > +
> > +using namespace libcamera::ipa;
> > +
> > +#define ASSERT_EQ(a, b) \
> > +if ((a) != (b)) { \
> > + std::cout << #a " != " #b << " (line " << __LINE__ << ")" \
> > + << std::endl; \
> > + return TestFail; \
> > +}
> > +
> > +class VectorTest : public Test
> > +{
> > +protected:
> > + int run()
> > + {
> > + Vector<double, 3> v1;
>
> My compiler (g++ 14.2.1) complains that
>
> ../test/ipa/libipa/vector.cpp: In member function ‘virtual int VectorTest::run()’:
> ../test/ipa/libipa/vector.cpp:18:9: error: ‘v1’ is used uninitialized [-Werror=uninitialized]
> 18 | if ((a) != (b)) { \
> ../test/ipa/libipa/vector.cpp:31:17: note: in expansion of macro ‘ASSERT_EQ’
> 31 | ASSERT_EQ(v1[0], 0.0);
> | ^~~~~~~~~
> ../test/ipa/libipa/vector.cpp:29:35: note: ‘v1’ declared here
> 29 | Vector<double, 3> v1;
> | ^~
>
> Automatic initialization is apparently the point of the test, so what's wrong?
I didn't notice that with gcc 13 initially, but I've now been able to
reproduce the issue.
The default constructor indeed doesn't initialize the memory. This could
be changed to initialize all entries to T{}, or we could do so
explicitly in callers where it matters.
> > + ASSERT_EQ(v1[0], 0.0);
> > + ASSERT_EQ(v1[1], 0.0);
> > + ASSERT_EQ(v1[2], 0.0);
> > +
> > + ASSERT_EQ(v1.length(), 0.0);
> > + ASSERT_EQ(v1.length2(), 0.0);
> > +
> > + Vector<double, 3> v2{{ 1.0, 4.0, 8.0 }};
> > +
> > + ASSERT_EQ(v2[0], 1.0);
> > + ASSERT_EQ(v2[1], 4.0);
> > + ASSERT_EQ(v2[2], 8.0);
> > +
> > + ASSERT_EQ(v2.x(), 1.0);
> > + ASSERT_EQ(v2.y(), 4.0);
> > + ASSERT_EQ(v2.z(), 8.0);
> > +
> > + ASSERT_EQ(v2.r(), 1.0);
> > + ASSERT_EQ(v2.g(), 4.0);
> > + ASSERT_EQ(v2.b(), 8.0);
> > +
> > + ASSERT_EQ(v2.length2(), 81.0);
> > + ASSERT_EQ(v2.length(), 9.0);
> > + ASSERT_EQ(v2.sum(), 13.0);
> > +
> > + Vector<double, 3> v3{ v2 };
> > +
> > + ASSERT_EQ(v2, v3);
> > +
> > + v3 = Vector<double, 3>{{ 4.0, 4.0, 4.0 }};
> > +
> > + ASSERT_EQ(v2 + v3, (Vector<double, 3>{{ 5.0, 8.0, 12.0 }}));
> > + ASSERT_EQ(v2 + 4.0, (Vector<double, 3>{{ 5.0, 8.0, 12.0 }}));
> > + ASSERT_EQ(v2 - v3, (Vector<double, 3>{{ -3.0, 0.0, 4.0 }}));
> > + ASSERT_EQ(v2 - 4.0, (Vector<double, 3>{{ -3.0, 0.0, 4.0 }}));
> > + ASSERT_EQ(v2 * v3, (Vector<double, 3>{{ 4.0, 16.0, 32.0 }}));
> > + ASSERT_EQ(v2 * 4.0, (Vector<double, 3>{{ 4.0, 16.0, 32.0 }}));
> > + ASSERT_EQ(v2 / v3, (Vector<double, 3>{{ 0.25, 1.0, 2.0 }}));
> > + ASSERT_EQ(v2 / 4.0, (Vector<double, 3>{{ 0.25, 1.0, 2.0 }}));
> > +
> > + ASSERT_EQ(v2.min(v3), (Vector<double, 3>{{ 1.0, 4.0, 4.0 }}));
> > + ASSERT_EQ(v2.min(4.0), (Vector<double, 3>{{ 1.0, 4.0, 4.0 }}));
> > + ASSERT_EQ(v2.max(v3), (Vector<double, 3>{{ 4.0, 4.0, 8.0 }}));
> > + ASSERT_EQ(v2.max(4.0), (Vector<double, 3>{{ 4.0, 4.0, 8.0 }}));
> > +
> > + ASSERT_EQ(v2.dot(v3), 52.0);
> > +
> > + v2 += v3;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 5.0, 8.0, 12.0 }}));
> > + v2 -= v3;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 1.0, 4.0, 8.0 }}));
> > + v2 *= v3;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 4.0, 16.0, 32.0 }}));
> > + v2 /= v3;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 1.0, 4.0, 8.0 }}));
> > +
> > + v2 += 4.0;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 5.0, 8.0, 12.0 }}));
> > + v2 -= 4.0;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 1.0, 4.0, 8.0 }}));
> > + v2 *= 4.0;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 4.0, 16.0, 32.0 }}));
> > + v2 /= 4.0;
> > + ASSERT_EQ(v2, (Vector<double, 3>{{ 1.0, 4.0, 8.0 }}));
> > +
> > + return TestPass;
> > + }
> > +};
> > +
> > +TEST_REGISTER(VectorTest)
--
Regards,
Laurent Pinchart
More information about the libcamera-devel
mailing list