The new test environment from Tizen is the Web-TCT test suite. This was written for testing web components, but can easily be used for testing Dali.
-Each of the DALi repositories, **dali-core**, **dali-adaptor** and **dali-toolkit**, have their own test suites under the `automated-tests` folder. Within the src folder are a number of secondary folders - these correspond to 'managed' tests (for testing the frozen API for products), unmanaged tests (tests for other APIs) and internal (for desktop testing only)
+Each of the DALi repositories, **dali-core**, **dali-adaptor** and **dali-toolkit**, have their own test suites under the `automated-tests` folder. Within the src folder are a number of secondary folders - these correspond to 'API' tests and internal (for desktop testing only)
Installation
------------
export CC=gcc
export CXX=g++
git clean -fxd . # Only do this in the build folder
- autoreconf --install
- CXXFLAGS='-g -O0 --coverage' LDFLAGS='--coverage' ./configure --prefix=$DESKTOP_PREFIX --enable-debug
+ CXXFLAGS='-g -O0 --coverage' LDFLAGS='--coverage' cmake -DCMAKE_INSTALL_PREFIX=$DESKTOP_PREFIX -DCMAKE_BUILD_TYPE=Debug
make -j8 install
Note, you __must__ use a local build and not a distributed build, and you __must__ also build with debug enabled to allow *DALI_ASSERT_DEBUG* to trigger on wrong behaviour ( Which should always be a test case failure! )
+Further note that, for the following, your gcov version must match the version of the compiler.
+
Building the tests
------------------
Executing the tests
-------------------
+To see a list of all of the options:
+
+ ./execute.sh -h
+
To execute tests, cd into automated-tests and run
./execute.sh
-This will execute dali, dali-unmanaged and dali-internal test sets. Note that the output summary for the first will be printed before running the second.
+This will execute dali and dali-internal test sets. Note that the output summary for the first will be printed before running the second.
-To execute a subset of tests, you can run individual test sets, e.g.
+By default the tests execute in parallel, which is faster but does not produce any test case output files. Use this to execute the tests in series and log test output to stdout/err
- ./execute.sh dali-adaptor
+ ./execute.sh -S
-or for finer granularity, specify a test source file:
+To use test kit lite, (which is very slow),
- ./execute.sh dali-adaptor-internal utc-Dali-GifLoader.cpp
+ ./execute.sh -s
-To see the results, you can open the summary.xml in firefox.
+To see the test kit lite results, copy the style folder from web-tct_2.2.1_r1/tools/tct-mgr/style into automated-tests and run
firefox --new-window summary.xml
-To get coverage output, run
+To execute a subset of tests, you can run individual test sets, e.g.
+
+ ./execute.sh dali-adaptor
+
+To get full coverage output (you need to first build dali libraries with
+--coverage), run
./coverage.sh
+To check the coverage of your patch, (the build server uses its own copy
+of these scripts), you can use
+
+ ./patch-coverage.pl -q [diff-spec]
+
+to get a summary, or
+
+ ./patch-coverage.pl [diff-spec]
+
+to get textual output, or
+
+ ./patch-coverage.pl -o out.html [diff-spec]
+
+to get HTML output (used by build server).
+
+diff-spec is any refspec accepted by git-diff. If it's left out, it creates
+a refspec to the latest commit, or uses the index/working tree.
+
Testing on target
=================
-To build for target, first build and install dali-core, dali-adaptor and dali-toolkit, then build dali-capi without --keep-packs option.
+To build for target, first build and install dali-core, dali-adaptor and dali-toolkit.
You will need to install libconfig-tiny-perl:
tct-mgr
-This will bring up the java test suite program. You should see the Plan pane with a list of all tests in. Select the tct-dali-core-tests and tct-dali-unmanaged-tests. and you will be offered a dialog to choose a test plan: either create a new one or use temp.
+This will bring up the java test suite program. You should see the Plan pane with a list of all tests in. Select the tct-dali-core-tests. and you will be offered a dialog to choose a test plan: either create a new one or use temp.
Select dali test suite, and click Run, then "Create a new plan", and call it "Dali-Core" or some such. It will now run the dali-test suite.
You can find the output files under /opt/tct/manager/result/
If you are adding test cases for new or existing managed API (CAPI), you need to add your changes to the src/dali mirror, and copy your change to the managed test suite in core-api. You need to inform HQ of your update.
-To Unmanaged API
-----------------
-
-If you are adding test cases for unmanaged API, you need to add your changes to the src/dali-adaptor-unmanaged test suite.
-
For internal API
----------------
END_TEST;
}
-
-Note that **there must be no extra whitespace in the method signature** (i.e., it must violate our coding convention and follow __exactly__ this pattern: `int UtcDaliMyTestcaseName(void)`), as it's parsed by an awk script to auto-generate the testcase arrays in the main header file.
+Note that **the parentheses in the method signature must not be empty** (i.e., it must violate our coding convention and follow __exactly__ this pattern: `int UtcDaliMyTestcaseName(void)`), as it's parsed by an awk script to auto-generate the testcase arrays in the main header file. Neither may any comments on the same line contain empty parentheses.
You can contine to use the TET api, e.g. `tet_infoline`, `tet_result` and our test check methods `DALI_TEST_CHECK`, `DALI_TEST_EQUALS`, etc.
If you are adding new test files, then you need to add the filename to the SET(TC_SOURCES...
section of CMakeLists.txt (this is also parsed by an awk script prior to building)
+Good Practices
+--------------
+Use DALI_TEST_EQUALS to test actual value against expected value, like this:
+
+ DALI_TEST_EQUALS( actor.GetProperty< float >( Actor::Property::COLOR_ALPHA ), 0.9f, TEST_LOCATION );
+
+This will speed up debugging in case the test some day fails. There is also a variant to test that value is greater than expected:
+
+ DALI_TEST_GREATER( textureBindIndex[1], textureBindIndex[2], TEST_LOCATION );
+
+When doing negative tests where your code uses DALI_ASSERT_ALWAYS, use the DALI_TEST_ASSERTION macro, like below:
+
+ DALI_TEST_ASSERTION(
+ {
+ animation.AnimateTo( Property( actor, Actor::Property::PARENT_ORIGIN ), targetParentOrigin );
+ }, "IsPropertyAnimatable( index )" );
+
+This macro will catch the DALi Exception and check that the correct assert message was included. It will also fail the test in case the assert did not occur. It also reduces the amount of false positive error logging whilst the is being thrown making it easier to see the real errors.
+
+Note, DALI_ASSERT_DEBUG cannot be tested as tests execute against release version of the code.
+
+Use additional scope to control the life of stack allocated objects, such as DALi handles
+
+ // try reparenting an orphaned child
+ {
+ Actor temporaryParent = Actor::New();
+ temporaryParent.Add( child );
+ DALI_TEST_EQUALS( parent2.GetChildCount(), 0u, TEST_LOCATION );
+ }
+ // temporaryParent has now died, reparent the orphaned child
+ parent2.Add( child );
+ DALI_TEST_EQUALS( parent2.GetChildCount(), 1u, TEST_LOCATION );
+
+Always test the output of your test by making your code fail!!!
+
Debugging
=========
On desktop, you can debug the tests by running gdb on the test program:
$ cd automated-tests
- $ gdb build/src/dali/tct-dali-core
+ $ gdb build/src/dali-adaptor/tct-dali-adaptor-core
gdb> r <TestCase>
replace `<TestCase>` with the name of the failing testcase.
projects / platform / core / uifw / dali-adaptor.git / blobdiff