Debugging Test Problems |
Previous | Next | Contents |
There are a number of reasons that tests can fail to execute properly. This chapter provides some approaches for dealing with these failures. Please note that most of these suggestions are only relevant when running the test harness in GUI mode.
This chapter includes the following topics:
The goal of a test run is for all tests in the test suite that are not filtered out to have passing results. If the root test suite folder contains tests with errors or failing results, you must troubleshoot and correct the cause to satisfactorily complete the test run.
Errors: Tests with errors could not be executed by the JavaTest harness. These errors usually occur because the test environment is not properly configured.
Failures: Tests that fail were executed but had failing results.
The Test Manager GUI provides you with a number of tools for effectively troubleshooting a test run. See the JavaTest User’s Guide and JavaTest online help for detailed descriptions of the tools described in this chapter. Ant test execution tasks provide command-line users with immediate test execution feedback to the display. Available JTR report files and log files can also help command-line users troubleshoot test run problems.
For every test run, the JavaTest harness creates a set of report files
in the reports directory, which you specified by setting the
report.dir
property in the <TS_HOME>/lib/jaxb_tck23.jte
file. The report files contain
information about the test description, environment, messages,
properties used by the test, status of the test, and test result. After
a test run is completed, the JavaTest harness writes HTML reports for
the test run. You can view these files in the JavaTest ReportBrowser
when running in GUI mode, or in the Web browser of your choice outside
the JavaTest interface. To see all of the HTML report files, enter the
URL of the report.html
file. This file is the root file that links to
all of the other HTML reports.
The JavaTest harness also creates a summary.txt
file in the report
directory that you can open in any text editor. The summary.txt
file
contains a list of all tests that were run, their test results, and
their status messages.
The work directory, which you specified by setting the work.dir
property in the <TS_HOME>/lib/jaxb_tck23.jte
file, contains several files that were
deposited there during test execution: harness.trace
, log.txt
,
lastRun.txt
, and testsuite
. Most of these files provide information
about the harness and environment in which the tests were executed.
Note
|
You can set |
If a large number of tests failed, you should read Configuration Failures to see if a configuration issue is the cause of the failures.
Use the test tree in the JavaTest GUI to identify specific folders and tests that had errors or failing results. Color codes are used to indicate status as follows:
Green: Passed
Blue: Test Error
Red: Failed to pass test
White: Test not run
Gray: Test filtered out (not run)
Click a folder in the test tree in the JavaTest GUI to display its tabs.
Choose the Error and the Failed tabs to view the lists of all tests in and under a folder that were not successfully run. You can double-click a test in the lists to view its test information.
To display information about a test in the JavaTest GUI, click its icon in the test tree or double-click its name in a folder status tab. The tab contains detailed information about the test run and, at the bottom of the window, a brief status message identifying the type of failure or error. This message may be sufficient for you to identify the cause of the error or failure.
If you need more information to identify the cause of the error or failure, use the following tabs listed in order of importance:
Test Run Messages contains a Message list and a Message section that display the messages produced during the test run.
Test Run Details contains a two-column table of name/value pairs recorded when the test was run.
Configuration contains a two-column table of the test environment name/value pairs derived from the configuration data actually used to run the test.
Note
|
You can set |
Report files are another good source of troubleshooting information. You may view the individual test results of a batch run in the JavaTest Summary window, but there are also a wide range of HTML report files that you can view in the JavaTest ReportBrowser or in the external browser or your choice following a test run. See Section 5.5, "Test Reports," for more information.
Configuration failures are easily recognized because many tests fail the
same way. When all your tests begin to fail, you may want to stop the
run immediately and start viewing individual test output. However, in
the case of full-scale launching problems where no tests are actually
processed, report files are usually not created (though sometimes a
small harness.trace
file in the report directory is written).
Here are few tips to verify the errors:
Check your .jti
file settings in the Configuration Editor. You may
have launched the JavaTest harness with prior status set to any
instead of ignore
.
Check if the env.html
report file is generated, else, the JavaTest
GUI provides the ability to view evaluated variables used by the tests.
Select Configure and then Show Test Environment from the Test
manager menu to view the contents of the test environment.
Verify if the Configuration Editor has generated a configuration question log when you complete a configuration interview. You can use the configuration question log to review your answers to all the questions in the current configuration. Select Configure and Show Question Log from the Test Manager menu to view the current configuration interview.
You can view the properties of a test manager by selecting View and then Properties from the Test Manager menu. The Test Manager Properties dialog box contains Test Suite, Work Directory, Configuration, and Plugin information.
See JavaTest User’s Guide , Graphical User Interface, or JavaTest online help for a detailed description of the Test Manager Properties dialog box.
If the JavaTest harness detects test suite errors, then it displays an advisory dialog box. You can view detailed information about the test suite errors by selecting View and then Test Suite Errors from the Test Manager menu.
See JavaTest User’s Guide , Graphical User Interface, or JavaTest online help for a detailed description of the Test Manager: Test Suite Errors dialog box.
To better understand and debug the test failures, this section describes how the tests are selected for a test run and how they are then executed.
Test Execution results are reported as one of the three states:
Pass: A test passes when the functionality being tested behaves as expected. All tests are expected to pass.
Fail: A test fails when the functionality being tested does not behave as expected.
Error: A test is considered to cause error when something (usually a configuration problem) keeps the test from being executed as expected. Errors often indicate a systemic problem - a single configuration problem can cause many tests to fail. For example, if the path to the Java runtime is configured incorrectly, then no tests can run and will result in an error.
Prior to the start of a test run, the JavaTest harness selects tests for the run based on the following factors:
Tests to be Run: The JavaTest harness finds tests listed in the Tests to be Run field of the JavaTest configuration. You can specify the sub-branches of the tree as a way of limiting the tests that are executed during a test run. The JavaTest harness browses through the tree starting with its sub-branches or tests you specify, and executes all tests that it finds.
Exclude List: Tests listed in the appropriate exclude list are deselected prior to the start of a test run. For details about exclude lists and their role in the certification process, see Procedure for Certification.
Keywords: A test can be selected based on keywords specified in the Keywords field in the test description. The possible keywords are the following: schema, document, runtime, positive, negative, bindinfo, jaxb_not_required, java_to_schema.
Prior Status: Use the combo box and check boxes to select tests in a test run based on their outcome on a prior test run. Prior status is evaluated on a test-bytest basis using information stored in result files (.jtr) written in the work directory.
Based on the test keywords and the executeClass
parameter of the test
description, all the tests are divided in four execution models:
Schema Tests: The keyword schema indicates this test execution model. The
schema tests are the Schema Compiler Tests. These tests have a schema file
(*.xsd) in the test description parameter source. The schema is compiled to the
work directory, and the output directory option of the XML Binding schema compiler is
used.
The package specified in the test description parameter package is used
to set the output package parameter to the schema compiler. If the test
has the keyword negative, then the schema compiler must report an error
compiling the test schema. Otherwise, the schema is positive and may
have XML Binding custom binding information if the keyword bindinfo is
present. + The positive schema is expected to compile successfully.
After the compilation passes, the signature test (the class specified
in the test description parameter executeClass) runs with arguments
specified in the test description parameter executeArgs.
Document Tests: The keyword document along with the class JAXBTest
specified in the test description parameter executeClass indicate this
test execution model. The document tests are the subset of the Content
Tree Tests, with the exception of the content tree modification stage.
The last two stages are applicable only to valid XML documents. The
first stage (schema compilation) is performed as for the schema tests.
All stages except the first one are performed by executing appropriate
test cases (unmarshal, compareContent, and marshal respectively) of the
class specified in test description parameter executeClass, with
arguments specified in the test description parameter executeArgs.
All parameter entries starting with $ are resolved in the
environment, but no errors occur if any of these values are not
defined. If the exclude list identifies test cases to be excluded, then
these cases are added to the test arguments using the -exclude option.
Note that all schemas used in the document tests are valid and the
tests may not have the keyword negative. Invalid documents are
indicated by option -invalid in the parameter executeClass.
XML Binding Framework API Tests: The XML Binding framework API tests have their own
classes (other than the signature test or JAXBTest) specified in the
test description parameter executeClass. The tests may or may not have
schema file specified in parameter schema.
First, if the schema is specified, it is compiled as described for the
schema tests. Then the test executes the class specified in test
description parameter executeClass with arguments specified in the test
description parameter executeArgs.
All parameter entries starting with $ are resolved in the
environment, but no errors occur if any of these values are not
defined. If the exclude list identifies test cases to be excluded, then
these cases are added to the test arguments using the -exclude option.
Java -to-Schema Tests: The keyword /java_to_schema/ indicates this test execution model. The Java-to-Schema Tests are the same as described in Schema Compiler Tests. These tests have a Java files (*.java) in the test description parameter source. The generated schema is placed in the work directory, and the output directory option of the XML Binding schema generator is used. If the test has the attribute negative, then the schema generator must report an error processing the test Java files. Otherwise, the Java code is correctly annotated and should be processed successfully. After the generation passes, the validation test (the class specified in the test description parameter /executeClass/) runs with arguments specified in the test description parameter /executeArgs/. The validation test checks that a valid XML document, taken from the test description parameter /executeArgs/ follows the just-generated schema, and fails to check invalid XML documents.
XML Validation Tests: XML Validation can be implemented through JAXP or some other mechanism. If the validation is implemented through JAXP, then validation checking tests can be skipped by deselecting them in the test interview. If you use some other tools for schema validation, then XML validation tests are mandatory.
Previous | Next | Contents |