Release notes
Release notes for TPTP V4.1 (all releases)

General Notices
1.0 Known problems and limitations
1.1 Generic Log Adapter
1.1.1 Problems running Generic Log Adapter using IBM's Java Runtime Environment (JRE) v1.4.0 and v1.4.1
1.1.2 Importing a log file from a remote z/OS system may result in incomplete data
1.1.3 Generic Log Adapter does not support creating rules to parse multiple timestamp formats
1.1.4 Formatter errors in Problems view when new Generic Log Adapter adapter file is run in the Generic Log Adapter perspective
1.1.5 Generic Log Adapter Documentation Error - Configuring the parser component for a rules based adapter
1.1.6 Not all Generic Log Adapter messages are displayed in the Problems view when running an adapter file in the Generic Log Adapter perspective
1.1.7 Parsing a log file that requires a converter command executed on AIX sometimes fails
1.1.8 Generic Log Adapter does not create child ExtendedDataElements correctly
1.1.9 No error message when Logging Agent Outputter fails to communicate with Agent Controller
1.2 Agent Controller
1.2.1 Old Agent Controller and Integrated Agent Controller PATH settings may result in unexpected behavior in the Profiling and Logging and Test perspectives
1.2.2 Profiling fails when Agent Controller and Integrated Agent Controller run concurrently
1.2.3 Attach Java Process does not refresh the agent list correctly
1.2.4 Problems connecting, profiling, or testing on Linux with the Local Direct Connection
1.2.5 The Local Direct Connection is used by one workbench exclusively
1.2.6 Launching a profiling or test application may fail through a client side firewall
1.2.7 Agent Controller uninstall instructions are missing a step
1.2.8 Using the Multiplexed Control Channel for data transmission to the client fails for large records
1.3 Log and Trace Analyzer
1.3.1 Entity expansion limit exception when importing a large log file
1.3.2 Simultaneous clients access using the same database account is not supported
1.3.3 Resource path collision
1.3.4 Improving performance of large log support
1.3.5 Restart DB2 after creating database and tables
1.3.6 Intermittent SQL exception
1.3.7 Errors reported when creating or dropping Cloudscape tables for large log support
1.3.8 The Log View and Log Interactions View do not refresh when a resource is deleted
1.3.9 Attempts to view Correlation by Time for log records fails when the Distributed Layout is selected
1.4 Probekit
1.4.1 Probekit may fail to instrument certain large Java methods

1.5 Profiling Tool
1.5.1 Cancel button does not cancel filter set modifications
1.5.2 Method counts incorrect with JIT Inlining
1.5.3 Method level CPU time statistics limitations on AIX and Solaris
1.5.4 Code Coverage Statistics view does not display empty, unreferenced methods
1.5.5 With the Sun JVM, some method calls are not traced
1.5.6 Execution History - Full Graphical Details profiling set involves collection of boundary classes

1.5.7 Open extension point description option fails
1.5.8 The Profiling Monitor view does not display profiling type indicators for profiling files generated on z/OS
1.5.9 The Collect Object References option returns a launch error on z/OS
1.5.10 Problem with garbage collection when using IBM JDK 1.4.1
1.5.11 Profiling on Solaris using the Sun JDK 1.4.x or on HP using the HP JDK 1.4.x may cause JVM to crash
1.5.12 Potential crash when running in standalone mode with STACK_INFORMATION=contiguous on Solaris
1.5.13 Negative timeout values for WAIT and WAITED events
1.5.14 Incorrect monitor dumps with IBM JDK 1.4.2
1.6 Statistical Console
1.7 Test
1.7.1 Common Test Issues
1.7.1.1 National characters are currently not supported in creating test suites and test cases
1.7.2 JUnit plugin tests to TPTP import option not documented

1.7.3 Limitations associated with JUnit Plug-in Testing
1.7.4 Test execution may hang due to &quote; characters in the org.eclipse.hyades.test.tools.core_4.1.0 configuration
1.7.5 SWT error when running the Manual Test View on Linux
1.7.6 Automated GUI Recorder test case with position-based recording enabled may not capture all mouse/keyword events
1.7.7 Automated GUI Recorder test log view displays each iteration of the loop as separate entry
1.7.8 SWT error when running Manual Test View on Linux on a multiprocessor machine
1.7.9 Test execution fails on some Linux platforms using Sun JRE 1.5
1.8 Charting Service
1.8.1 SVG filter effects are not supported in Mozilla native SVG browsers
1.8.2 The Adobe SVG v3.0 plug-in is required to view test reports in Internet Explorer
1.8.3 Hungarian is not supported by SVG Adobe Viewer 3.0
1.8.4 The Adobe Acrobat v5.0 installation program installs an older version of the Adobe SVG viewer
1.8.5 SWT charts are distorted when using the org.eclipse.tptp.platform.report.core.provisional.Generator.run method

General Notices

References to "TPTP" should be read as equivalent to "Test and Performance Tools Platform".

Any references to "TPTP" in the documentation, installation, migration, or readme files should be read as equivalent to "Test and Performance Tools Platform".

List of supported operating system and JREs

The entire list can be found here: TPTP v4.1 OS and JRE support

And the list of only the deployment operating systems and JREs is found here: TPTP v4.1 OS and JRE deployment support

1.0 Known problems and limitations

1.1 Generic Log Adapter

1.1.1 Problems running Generic Log Adapter using IBM's Java Runtime Environment (JRE) v1.4.0 and v1.4.1

When executing Generic Log Adapter with older versions of IBM JDK (1.4.0 and 1.4.1) error messages like the following may be issued in the hgla.log file (in the msg attribute of CommonBaseEvent records).

IWAT0383E Adapter Validation Error: TargetNamespace.1: Expecting namespace
'http://www.eclipse.org/hyades/schema/commonbaseevent1_0_1.xsd', but
the target namespace of the schema document is
'http://www.ibm.com/AC/commonbaseevent1_0_1'

IWAT0321E Regular expression is malformed.  Exception = Unexpected character 
'?' Pattern = ^([^\d{1,3}][^\.][^\d{1,3}].*)  Index = 5

When using IBM's Java Runtime Environment (JRE) to execute the Generic Log Adapter or import log files into the Log and Trace Analyzer, it is recommended to use IBM's Java Runtime Environment (JRE) v1.4.1 Service Release (SR2) and above or IBM's Java Runtime Environment (JRE) v1.4.2.

1.1.2 Importing a log file from a remote z/OS system may result in incomplete data

Bugzilla defect: 80730

Parsing a log file using Generic Log Adapter on a z/OS system may result in incomplete data. The parsing operation may stop prematurely and not all of the log records are parsed. This problem occurs when one of the following IBM JDK versions is installed on the z/OS system:

1.4.1 PTF UQ88783
1.4.2 PTF UQ90449
1.4.2 PTF UQ93743 

This problem is fixed in IBM JDK 1.4.2 with PTF UK00802. Upgrade the JDK to that version or a later version. If you cannot upgrade the JDK version, execute Generic Log Adapter with the Java JIT compiler disabled to work around the problem. To disable the Java JIT compiler, add the following JVM argument to the java command in the script file used to execute Generic Log Adapter:

-Djava.compiler=NONE

1.1.3 Generic Log Adapter does not support creating rules to parse multiple timestamp formats

Generic Log Adapter does not support parsing log files that have locale sensitive timestamp formats with a single rules-based adapter configuration file. If an application generates log files containing timestamps that have formats dependant on the locale it is generated in, then these logs cannot be parsed with a single rules-based adapter. For example, if the date format is MM/dd/yy in log files generated on en_US systems, yy/MM/dd in log files generated on ja_JP systems, and dd.MM.yy for log files generated on de_DE systems then a separate adapter configuration file is required to parse each log file, each one having a parsing rule with the correct timestamp format for the locale.

1.1.4 Formatter errors in Problems view when new Generic Log Adapter adapter file is run in the Generic Log Adapter perspective

Bugzilla defect: 62103

The Problems view in the Generic Log Adapter perspective returns the following error when trying to execute a new Generic Log Adapter adapter file by clicking the Rerun adapter ... button:

IWAT0438E Common Base Event formatter N76D20B0042411D98000E0362B33D6F0 cannot 
create a CommmonBaseEvent because required property sourceComponentId is 
missing.

This message indicates that the formatter component of Generic Log Adapter could not create a Common Base Event because sourceComponentId is a required property of Common Base Event and it is missing. To work around this problem, add parser rules to the adapter file for the sourceComponentId attributes. Note that the situation property is also a required Common Base Event property. To avoid similar errors, add parser rules to the adapter for the situation property. Generic Log Adapter will only create CommonBaseEvents that contain all of the required properties.

1.1.5 Generic Log Adapter Documentation Error - Configuring the parser component for a rules based adapter

Bugzilla defect: 115357

In the online help for the Generic Log Adapter, information in the following help file is incorrect: Generic Log Adapter > Creating a log parser > Create a rules-based adapter > Configuring the parser component. In Step 3 under Adding and configuring attribute substitution rules, the sentence, "if you just want to assign the value of the warning to the Common Base Event msg attribute, specify the substitution as $1.", should state "if you want to assign the part of the message after the dash to the Common Base Event msg attribute, specify the substitution as $2." The Substitute value in the accompanying screen shot should have a value of $2.

1.1.6 Not all Generic Log Adapter messages are displayed in the Problems view when running an adapter file in the Generic Log Adapter perspective

Bugzilla defect: 101367

Sometimes when the Generic Log Adapter configuration file is incorrect and it is executed in the Generic Log Adapter perspective by clicking the Rerun adapter ... button, not all of the error messages are shown in the Problems tab view. The number of errors indicated at the top of the Problems tab is correct but less than that number are shown in the Problems table. To work around this problem click the Filters... button on the Problems tab tool bar and click OK in the Filters dialog window that is displayed. The table in the Problems tab view will be refreshed with all of the error messages.

1.1.7 Parsing a log file that requires a converter command executed on AIX sometimes fails

Bugzilla defect: 104053

Adapter configuration files that use a converter command in the sensor component to perform some conversion on the log data before parsing it may not work correctly on AIX. The converter command may fail to execute on AIX and the log file cannot be parsed. For example if the converter command is a java command, the converter will fail on AIX. Converter commands that are shell scripts will be executed on AIX so commands that fail can be wrapped in shell scripts and the shell script can be specified as the converter command in the adapter file.

1.1.8 Generic Log Adapter does not create child ExtendedDataElements correctly

Bugzilla defect: 116447

Executing rules-based adapter configuration files that include rules for child ExtendedDataElements, with Generic Log Adapter, will produce CommonBaseEvents with incorrect child ExtendedDataElements. It is recommended not to create adapter files with rules for child ExentdedDataElements with this release of TPTP.

1.1.9 No error message when Logging Agent Outputter fails to communicate with Agent Controller

Bugzilla defect: 113171

When attempting to do live monitoring of a log file using standalone Generic Log Adapter configured with a Logging Agent Outputter (CBELogOutputter), there is no error message issued if the Agent Controller runtime DLL's are not accessible to the Generic Log Adapter process. The Generic Log Adapter process will hang waiting for the logging agent to register with Agent Controller and be monitored. To workaround this problem ensure the following:

On Windows systems, the Agent Controller bin directory must be included in the Windows PATH environment variable.

or

On non-Windows systems, the Agent Controller lib directory must be included in the LIBPATH, LD_LIBRARY_PATH, or SHLIB_PATH environment variable, depending on the operating system.

1.2 Agent Controller

1.2.1 Old Agent Controller and Integrated Agent Controller PATH settings may result in unexpected behavior in the Profiling and Logging and Test perspectives

Bugzilla defect: 113800

In the past, TPTP workbench users required the services of the Agent Controller to profile applications or run tests. Now, the Integrated Agent Controller allows users to profile Java applications locally and run TPTP tests locally without requiring the Agent Controller on the local machine.

In a standard Agent Controller installation, the Agent Controller's bin directory is added to the system PATH. When the Integrated Agent Controller is used for local profiling or tests, its bin directory is appended to the system PATH. If an Agent Controller bin directory appears before the Integrated Agent Controller entry in the system PATH, the Agent Controller libraries are used instead. This may result in unexpected behavior if an older Agent Controller is installed on the system.

If the Integrated Agent Controller is to be used for agent management, a suitable work around is to remove Agent Controller bin directories from your system path.

1.2.2 Profiling fails when Agent Controller and Integrated Agent Controller run concurrently

In this release, the workbench does not prevent a user from profiling using Integrated Agent Controller (by Attach - Java Process or External Java Application) when the Agent Controller is running. This will cause profiling to fail.

Make sure that Agent Controller is not running when profiling with Integrated Agent Controller. One way to check if it's safe to use the Integrated Agent Controller is by testing the connection before profiling. This can be accomplished by selecting Local Direct Connection in the Host tab and clicking the Test Connection button. If Agent Controller is running, you should see a warning indicating that fact and you should not use Integrated Agent Controller for profiling.

1.2.3 Attach Java Process does not refresh the agent list correctly

When attaching to a Java process during profiling, sometimes you will see the Retrieving list... message in the agent tab if the Refresh Data button is pressed. The system seems to be waiting for an updated agent list. You may also see the message Fetching children of host in the status area (at the bottom right) of the workbench.

If you wait until the agent list is updated, you may not get an accurate list. When you encounter this situation, press the Refresh Data button again to get the accurate agent list immediately. You can even press the button while the Retrieving list... message is still being displayed.

1.2.4 Problems connecting, profiling, or testing on Linux with the Local Direct Connection

A number of seemingly disparate symptoms are all due to the same causes. These problems occur only on Linux when using the Local Direct Connection (also known as the Integrated Agent Controller).

Symptom 1: Connection failed
An error dialog appears when you try to test or profile with the Local Direct Connection. The dialog says, "Connection failed on host localhost..."

Symptom 2: Launching a test case hangs
The status bar in the workbench's progress indicator hangs at 54% or 93% when launching a test case.

Symptom 3: Profiling data is not collected and the process console is unavailable
A launched process from the Profiling Perspective appears in the Profiling Monitor but no profiling data is collected. The process console standard in, standard out, and standard error are also unavailable.

These problems can be due to any of the following causes:

Cause 1: Native shared libraries required by Integrated Agent Controller need to be added to the LD_LIBRARY_PATH environment variable
To correct the problem, set the LD_LIBRARY_PATH before starting up Eclipse. Issue this command in a command shell:

export LD_LIBRARY_PATH=<Eclipse install directory>/plugins/org.eclipse.hyades.execution_4.1.0/os/linux/x86
Where <Eclipse install directory> is replaced with the fully-qualified path of the directory where you installed Eclipse.

Cause 2: The /tmp/IBMRAC/ramaster file needs to be deleted
When Agent Controller or Integrated Agent Controller terminate (either one or the other can be active at any time), the /tmp/IBMRAC/ramaster file should be deleted. If it isn't deleted, and neither controller is active, then delete it manually.

  1. See if Agent Controller is running by opening a command shell and typing ps -ae | grep RAServer. If there is at least one process ID listed then the Agent Controller is running. Terminate the Agent Controller process because the Agent Controller and Integrated Agent Controller are mutually exclusive. If it was running and you terminated it, then try testing or profiling again.
  2. If Agent Controller is not running, check to see if the file /tmp/IBMRAC/ramaster exists. If it exists, delete it, because it should exist only when Agent Controller is running. After deleting the file, try again.
Tip: If the Agent Controller is not running, then after shutting down Eclipse, check to see if the file /tmp/IBMRAC/ramaster is left behind. Ensure that it is deleted after Eclipse shuts down. Otherwise, during the next Eclipse workbench session, the Local Direct Connection cannot be started.

Cause 3: The JRE needs to be upgraded to a newer version.
If you are using IBM JDK 1.4.2, use at least Service Release 3 (java version cxia32142-20050929).

Cause 4: The JRE used to build a test case needs to be the JRE used to start Eclipse
Edit the Eclipse workbench preference:

  1. Click Window > Preference...
  2. Select Java > Installed JREs
  3. Select the JRE used to start Eclipse. If this JRE is not in the list, create it.

Cause 5: More memory needs to be allocated to run Integrated Agent Controller
Edit the eclipse.ini file in the Eclipse installation directory and change:

Cause 6: More file descriptors need to be allocated to run Integrated Agent Controller
Before starting up Eclipse,ssue this command in a command shell:

  ulimit -n 4096

1.2.5 The Local Direct Connection is used by one workbench exclusively

If you launch more than one Eclipse workbench, only one can use the Local Direct Connection (also known as the Integrated Agent Controller).

For example, say that you have two instances of the Eclipse workbench launched, and you want to profile an application from each workbench at the same time. Only one workbench can choose the Local Direct Connection in the Host tab of the Profile dialog. The other workbench, and any other workbenches opened subsequently, must use the local Agent Controller (e.g. localhost).

This limitation applies to all functionality in TPTP that uses an Agent Controller: monitoring, profiling, and testing.

1.2.6 Launching a profiling or test application may fail through a client side firewall

A client side firewall may cause communication between the Remote Agent Controller (RAC) and a profiling, logging, or test application to fail. When the RAC attempts to connect for the first time, firewall software may prompt the user to accept the connection. If the user response is delayed, the Agent Controller communication may timeout or fail. The application may not launch or the workbench viewer may not receive data. Subsequent attempts through the firewall usually succeed, as the firewall software no longer prompts for user input.

1.2.7 Agent Controller uninstall instructions are missing a step

Bugzilla defect: 117152

The Agent Controller getting_started.html files for non-Windows, non-OS/400 platforms are missing an uninstall step. The following final step should be performed:

Remove the <unzip directory>/lib directory from the system's LIBPATH or LD_LIBRARY_PATH or SHLIB_PATH environment variable.

This step should undo what was done in step 1 of Configuring and starting Agent Controller of getting_started.html.

1.2.8 Using the Multiplexed Control Channel for data transmission to the client fails for large records

Bugzilla defect: 115966

Configuring the Agent Controller to use the multiplexed control channel by setting isDataMultiplexed="true" in serviceconfig.xml will cause profiling or remote log import to end when a record larger than 8K bytes is recieved from Agent Controller. The profiling or remote log import operation will end and only records up to the large record will be shown in the views.

If you are collecting records larger then 8K from agents and they are being sent to the TPTP eclipse workbench via the Agent Controller then set isDataMultiplexed="false" in the Agent Controller serviceconfig.xml file.

1.3 Log and Trace Analyzer

1.3.1 Entity expansion limit exception when importing a large log file

If you import a large log file, the Import Log File wizard may fail with the following exception.

IWAT0353E Common Base Event logging agent outputter
AdapterStaticParserOutputterID1 cannot write to agent unknown due to
the following exception: Parser has reached the entity expansion
limit "64,000" set by the Application.

In particular Sun JVM versions, the entityExpansionLimit system property is set to constrain the total number of entity expansions when parsing XML documents. To solve this problem, set the entityExpansionLimit JVM option as follows:

-DentityExpansionLimit=500000

1.3.2 Simultaneous clients access using the same database account is not supported

Concurrent client access to a database might lock some tables. To unlock the tables, restart the Workbench and the database server.

1.3.3 Resource path collision

If you use the same database and database account for multiple workspaces, resource path collisions may result and resource will not be persisted into the database.

To work around this problem, use different project and monitor names in each workspace to avoid resource path collisions.

1.3.4 Improving performance of large log support

To improve the performance of large log support, run the following command after you have imported large logs or created correlations using large logs:

db2 -tvf plugins/org.eclipse.hyades.resources.database_x.x.x
             /scripts/runStatsForAllHyadesTablesDB2-8.1.sql

where x.x.x is the version number on the plug-in directory.

1.3.5 Restart DB2 after creating database and tables

After running the database and tables creation script, CreateDatabaseAndTablesDB2-8.1.sql, you must restart DB2. From a DB2 command window run db2stop and then db2start in order to apply the changes to the configuration parameters made by the script.

1.3.6 Intermittent SQL exception

Occasionally, the following SQL exception may occur:

com.ibm.db2.jcc.b.SQLException: NULLID.SYSSH203 0X5359534C564C3031

When this problem occurs, restart the workbench and the database, if using DB2 or Cloudscape in network mode. Logs imported when this exception occurs will not be valid and will have to be re-imported.

1.3.7 Errors reported when creating or dropping Cloudscape tables for large log support

Bugzilla defect: 116453

When creating or dropping the Cloudscape tables for large log support, because the same scripts support multiple levels of Cloudscape, error messages, such as missing jdbc drivers, are reported. These messages can be safely ignored.

1.3.8 The Log View and Log Interactions View do not refresh when a resource is deleted

Bugzilla defect: 117162

If you delete a log file or correlation in the Log Navigator view, the corresponding Log Interaction View and Log View may continue to appear populated. This problem only occurs if the Link with Viewer option is not selected. To work around this problem and refresh or clear the corresponding log views, select the Link with Viewer toolbar option in the Log Navigator view.

1.3.9 Attempts to view Correlation by Time for log records fails when the Distributed Layout is selected

Bugzilla defect: 117159

In the Log View, repeated attempts to view records correlated by time (accessible via the Correlation by Time pop-up menu option) fail and an empty UML2 Interactions view appears. Typically, the first Correlation by Time invocation succeeds and all subsequent attempts fail.

This problem only occurs when the Distributed Layout option is selected in the Log Navigator view. To work around this problem, select the Simply Layout option in the Log Navigator toolbar drop-down menu.

1.4 Probekit

1.4.1 Probekit may fail to instrument certain large Java methods

Bugzilla defect: 63336

The Probekit may fail to instrument certain large Java methods. If Probekit's byte- code insertion causes a method to grow so large that interior branches can no longer reach their targets, the Probekit will throw an exception and instrumentation will fail.

1.5 Profiling Tool

1.5.1 Cancel button does not cancel filter set modifications

The cancel button in the Filter Set page of the Edit Profiling Set wizard does not cancel filter set modifications. If you reopen the Filter Set page in the same launch configuration edit session, filter set changes appear as saved. To work around the problem and reset the filter set modifications, click Close to close the launch configuration dialog, instead of clicking Apply or Profile.

1.5.2 Method counts incorrect with JIT inlining

Bugzilla defect 70660 (closed as "Won't fix")

If you suspect that the method counts you are seeing in the analysis tools are too low, turn off JIT inlining, if you are using it. This problem happens only when with the IBM Java 2 Runtime Environment v.1.4.2, and only when JIT is enabled.

The only work-around for this problem is to turn off inlining. To do this, set the following environment variable:

JITC_COMPILEOPT=NINLINING

1.5.3 Method level CPU time statistics limitations on AIX and Solaris

In TPTP 3.0 and 4.0, method level CPU time statistics are available for collection. Optionally, you can view method level CPU time statistics in an additional column in the Method Statistics view or Method Invocation table. Platform limitations for this feature are as follows:

There is no support for method level CPU time statistics reporting on AIX 4.3.

On AIX Version 5.1, method level CPU time statistics reporting requires that the environment variable "AIXTHREAD_ENRUSG=ON" be exported.

The method level CPU time statistics feature is currently not supported on Solaris.

1.5.4 Code Coverage Statistics view does not display empty, unreferenced methods

The Java Virtual Machine Profiler Interface (JVMPI) does not include empty, unreferenced methods in profiled method listings. Accordingly, the Code Coverage Statistics view does not display methods which are empty and unreferenced.

1.5.5 With the Sun JVM, some method calls are not traced

Bugzilla defect: 69051

Using the Sun JDK on Windows, certain method calls in Java programs are not being traced by JVMPI.

There is no known workaround.

1.5.6 Execution History - Full Graphical Details profiling set involves collection of boundary classes

Bugzilla defect: 106793

The Execution History - Full Graphical Details profiling set (accessible via Run > Profile... > New > Profiling Tab > Overview Tab) differs significantly from the v3.2 Execution History profiling set.

In version 3.3 the original Execution History profiling set was renamed to Execution History - Statistical in order to facilitate addition of the new graphical profiling set called Execution History - Full Graphical Details. The Execution History - Statistical set should be used in place of Execution History in version 3.3 onwards.

The Execution History - Full Graphical Details set differs significantly from the original Execution History. Among other things, the Collect boundary classes excluded by the filter set option is selected by default in the graphical execution set, which involves the collection of extra classes and methods.

1.5.7 Open extension point description option fails

The Open extension point description option (accessible via the Extensions page of a sample project's plugin.xml) may fail with the following message. In this case, the extension is org.eclipse.hyades.ui.contexts.

Description for extension point "org.eclipse.hyades.ui.contexts"
cannot be found.

To work around this problem:

1.5.8 The Profiling Monitor view does not display profiling type indicators for profiling files generated on z/OS

Bugzilla defect: 116348

When you import a profiling file generated on z/OS, profling type indicators (for example, Basic Memory Analysis, Execution Time Analysis, and Method Code Coverage) do not appear listed below Agents in the Profiling Monitor view.

The profiling data, however, is imported successfully and the profiling views are available. To work around this problem and access a profiling view, right-click an Agent in the Profiling Monitor view, choose Open With >, and then select the profiling type.

1.5.9 The Collect Object References option returns a launch error on z/OS

Bugzilla defect: 117570

The following message may appear when you choose the Collect Object References option on z/OS. Note: You can invoke Collect Object References from the toolbar of the Profiling Monitor view or by right clicking on a profiling agent in the Profiling Monitor view.

IWAT0003E Problems occurred while launching the Java process.

IWAT0004W Problems occurred while launching the Java process
Reason: Heap dump failed because bad data passed from JVM.

This message only appears if both Basic Memory Analysis and Execution Time Analysis profiling data is collected, and the profiling mode is set to enabled when profiling an application from the command line (i.e. java -XrunpiAgent:server=enabled ...).

To work around the problem, retry the Collect Object References option. After closing the message dialog, a second (and any subsequent) Collect Object References invocations will succeed.

1.5.10 Problem with garbage collection when using IBM JDK 1.4.1

Bugzilla defect: 56182

If the user's application uses an extremely large amount of heap space, requesting Collect Object References or Run GC, can potentially cause the JVM to crash with the following error message:

 **Out of memory, aborting**

 *** panic: JVMCI023: Cannot allocate memory to collect heap dump in jvmpi_heap_dump
 abnormal program termination

You can try to work around this by running without the -Xmx parameter, if you are currently running with it. This problem has been detected on Windows and z/OS.

1.5.11 Profiling on Solaris using the Sun JDK 1.4.x or on HP using the HP JDK 1.4.x may cause JVM to crash

Bugzilla defect: 56404

Profiling on Solaris using the Sun JDK 1.4.x or on HP using the HP JDK 1.4.x may cause the JVM to crash.

The problem is due to a bug in the JVM. To work around this problem, use only one of the following profiling sets:

The problem arises if you use these profiling sets in combination or if "Show instance level" information is turned on.

Alternatively, you can upgrade to the JDK 1.4.2_08-b03 build on Sun or the JDK 1.4.2_04 build on HP, both of which fix the problem.

1.5.12 Potential crash when running in standalone mode with STACK_INFORMATION=contiguous on Solaris

Bugzilla defect: 50090

When profiling on Solaris, you may encounter problems with standalone profiling. The problem only occurs when STACK_INFORMATION=contiguous (or boundaryAndContiguous) and TRACE_MODE=full. This problem may result in your JVM crashing.

To work around this problem with STACK_INFORMATION=contiguous, set TRACE_MODE=noObjectCorrelation. The problem does not occur when STACK_INFORMATION=none or STACK_INFORMATION=normal.

1.5.13 Negative timeout values for WAIT and WAITED events

Bugzilla defect: 63969

When running with IBM 1.4.2 JDK, with the jvmpi profile option 'MONITOR_MODE=all' (in standalone mode), you may see negative timeout attributes on monitorWait and monitorWaited elements in their trace. These are actually extremely large timeout values cast as positive 64 bit integers. This bug is the result of a JDK bug.

The JDK bug has been fixed as of IBM JDK 1.4.2 SR1a. A solution is to upgrade to this JDK level or later.

1.5.14 Incorrect monitor dumps with IBM JDK 1.4.2

Bugzilla defect: 65193 and 72180

Because of a JDK bug, when running the Test and Performance Platform in standalone mode with the jvmpi profile option 'MONITOR_MODE=all', you may get incorrect monitor dumps. For bug 65193 particularly, this happens when the '-Xj9' VM argument is used.

The JDK bug has been fixed as of IBM JDK 1.4.2 SR1 (with -Xj9) and IBM JDK 1.4.2 SR3 (without -Xj9). A solution is to upgrade to these JDK levels or later.

1.6 Statistical Console

N/A

1.7 Test

1.7.1 Common Test Issues

1.7.1.1 National characters are currently not supported in creating test suites and test cases

Bugzilla defect: 88029

National characters are currently not supported in creating test suites and test cases. The Agent Controller will not be able to handle those characters and the execution of the test will fail.

The workaround for this problem is to use standard ASCII characters when creating test suites and test cases.

1.7.2 JUnit plugin tests to TPTP import option not documented

The JUnit plugin tests to TPTP option in the Import dialog is not described in the on-line help documentation. The JUnit plugin tests to TPTP option allows you to import JUnit plug-in tests as TPTP JUnit plug-in tests.
Note: to import JUnit plug-in tests as standard TPTP JUnit tests (as opposed to TPTP JUnit plug-in tests), use the JUnit plugin tests to TPTP import option.

To import JUnit plug-in tests as standard TPTP JUnit tests:

  1. From the File menu, choose File > Import... The Import dialog appears.
  2. Select JUnit plugin tests to TPTP and click Next. The JUnit plugin tests to TPTP dialog appears, showing plug-in projects found in the current workspace.
  3. Select the tests to import and click Finish. The imported tests are now available for editing and running in the Test Navigator view.

1.7.3 Limitations associated with the JUnit Plug-in Testing

Currently, the following JUnit test limitations exist:

  1. Workbench location limitations
    In order to run a JUnit plug-in test suite on a different host, one needs to create a "Workbench Location". This workbench location points to a host and a directory where Eclipse is installed. To sucessfully run a JUnit Plug-in test with a remote workbench location, the Eclipse installation must meet the following criteria:
    • The remote Eclipse installation must contain all plug-ins selected in the Preferences > Plug-in Development > Target Platform > Plug-ins preferences page.
    • The plug-in versions of the remote Eclipse installation must match the plug-in versions of the target platform.
    • The remote Eclipse installation must consist of a single site and should not consist of multiple "eclipse" directories.
  2. Workbench location editor
    The workbench location property page (available when opening a workbench location) shows the plug-ins available on the target platform instead of the plug-ins available in the remote Eclipse installation.
  3. Workspace versus External plug-ins Any plug-in required by the test plug-in is automatically deployed and installed to the remote Eclipse installation, if it is not already installed. If a required plug-in exists both in the workspace (in a plug-in project) and on the remote Eclipse installation, the latter will take precedence. To deploy using the workspace version of the plug-in, unselect the remote plug-in using the workbench location editor.
  4. Performance issue when a test plug-in has many dependencies
    If a JUnit plugin test is defined in a plug-in that has many dependencies on workspace plug-ins, it may take a long time to run. This is due to the required workspace plug-ins being deployed to the remote Eclipse installation each time the test is executed (even if the "remote" Eclipse installation is actually local). To avoid this performance problem, one can export the required workspace plug-ins to the remote Eclipse installation prior to running tests. Note: the plug-ins found on the remote Eclipse installation take precedence over the workspace plug-ins, provided that they are enabled in the workspace location editor.
  5. Known issues related to execution
    • If both Run in UI thread and Allow the same instance to be reused options are enabled, the Eclipse instance hosting the tests will be unresponsive between test executions. This does not affect the validity of the tests.
    • When Allow the same instance to be reused is enabled, the launched Eclipse instance will never shut down automatically. To shut it down, one can either use the UI (if it was not launched in headless mode), kill the java process (if it was launched in headless mode, or if tests are to run in the UI thread), or run a test that shuts down the platform (for instance a test that invokes eclipse.swt.widgets.Display.getDefault().syncExec(new Runnable() { public void run() { org.eclipse.ui.PlatformUI.getWorkbench().close(); }});).

1.7.4 Test execution may hang due to &quot; characters in the org.eclipse.hyades.test.tools.core_4.1.0 configuration

Bugzilla defect: 114028

TPTP test execution may hang due to &quote; characters in pluginconfig.xml, located in the /plugins/org.eclipse.hyades.test.tools.core_4.1.0/config folder of your Agent Controller installation.If test execution suspends indefinitely, check the contents of the pluginconfig.xml <parameter> tag. Depending on your platform, one of the two forms below should fix the problem. If you are using z/OS, for instance, remove the &quot; characters and verify that the JAVA_PATH environment variable does not include internal spaces.

<Parameter position="prepend"
value="-Dhyades.test.JVM=&quot;%JAVA_PATH%&quot;"/>

                <Parameter position="prepend"
value="-Dhyades.test.JVM=%JAVA_PATH%"/>

Note: the issue is primarily found on z/OS where the &quot; characters cannot be present in the Agent Controller configuration in order for test execution to function.

1.7.5 SWT error when running the Manual Test View on Linux

Bugzilla defect: 114028

The Manual Test View uses the Eclipse Standard Widget Toolkit (SWT) Browser. When running the Manual Test View on Linux, the following SWT error is thrown if Mozilla v1.4 to v1.6 GTK2 is not available and configured on the system:

Unknown Mozilla path (MOZILLA_FIVE_HOME is not set)]

A workaround is provided at: http://www.eclipse.org/swt/faq.php#browserlinux

1.7.6 Automated GUI Recorder test case with position-based recording enabled may not capture all mouse/keyword events

Bugzilla defect: 110726

TPTP 4.1's Tech Preview Automated GUI Recorder allows you to record and playback user interactions under the Eclipse platform. For more information, see the Automated GUI Recorder user's guide.

An Automated GUI Recorder test case with position-based recording enabled may not capture all mouse/keyword events. In order to make position-based recording platform-independent, it relies on mouse and keyword events that are reported by SWT. There are some areas of the shell in which SWT will not report mouse or keyword events (for example, when user clicks a file menu). Furthermore, SWT will not report mouse/keyword events when a context menu is active. It is recommended for the user to use object-based recording where possible and only enable position-based recording for interacting with drawn objects displayed in views.

1.7.7 Automated GUI Recorder test log view displays each iteration of the loop as separate entry

Bugzilla defect: 109163

TPTP 4.1's Tech Preview Automated GUI Recorder allows you to record and playback user interactions under the Eclipse platform. For more information, see the Automated GUI Recorder user's guide.

When launching an Automated GUI Recorder test suite that contains loops in its behaviour page, the test log view will display each iteration of the loop as one loop entry. For example, a test suite containing a single loop with 5 iterations will appear in the test log view as 5 separate loops. The verdict for each loop entry in the test log view is still valid.

1.7.8 SWT error when running Manual Test View on Linux on a multiprocessor machine

Bugzilla defect: 115648
Related SWT bugzilla defect: 115822

When running the Manual Test View on Linux on a multiprocessor machine, the following SWT error is thrown:
No more handles [gtk_init_check() failed]
There is no workaround at this time (see SWT defect #115822). An alternative solution is to run the Manual Test View on Linux on a single processor machine.

1.7.9 Test execution fails on some Linux platforms using Sun JRE 1.5

Bugzilla bug: 116725

Test execution fails and exhibits unexpected behaviour on some Linux platforms using Sun JRE 1.5. Typically, the execution status stops at about 60% and the following message appears.

IWAT4039E The Test Execution Harness reported the following problems:
org.eclipse.core.runtime.CoreException

Note: this defect only occurs when using the Integrated Agent Controller and is due to the existence of certain non-standard environment variables.

To work around this problem,

1.8 Charting Service

1.8.1 SVG filter effects are not supported in Mozilla native SVG browsers

The current Mozilla native SVG enabled browsers do not fully support the SVG 1.0 specification. Specifically, filter effects that provide shading for 3-D horizontal, vertical, and stack bar charts are not supported. In additional the scatter chart highlighting effect is not functional in Mozilla native SVG enabled browsers.

1.8.2 The Adobe SVG v3.0 plug-in is required to view test reports in Internet Explorer.

The Adobe SVG v3.0 plug-in or a later version is required to view test reports in Internet Explorer. The Adobe SVG Viewer can be downloaded at the Adobe SVG Web site.

1.8.3 Hungarian is not supported by SVG Adobe Viewer 3.0

Bugzilla defect: 88786

The SVG charts generated by the charting service were tested with Adobe SVG Viewer 3.0. Adobe does not provide an SVG Viewer 3.0 that supports Hungarian. Go to http://www.adobe.com/svg/ to view supported languages.

1.8.4 The Adobe Acrobat v5.0 installation program installs an older version of the Adobe SVG viewer

The Adobe Acrobat Reader version 5.0 install includes Adobe SVG Viewer version 2.0. However, the SVG charts can only be viewed using version 3.0 of the Adobe SVG Viewer. To view the charts properly the Adobe 3.0 SVG viewer must be installed following the Adobe Acrobat Reader version 5.0 installation. Go to http://www.adobe.com/svg/ to download the Adobe 3.0 SVG viewer plugin.

1.8.5 SWT charts are distorted when using the org.eclipse.tptp.platform.report.core.provisional.Generator.run method

Bugzilla defect: 119304

To generate a SWT chart a developer would typically run the following code:
	//Create an instance of the SWT renderer
	IRenderer renderer = new SWTGraphicViewer(parent, SWT.DIALOG_TRIM );

	//show the chart in the SWT renderer
	Generator.instance().run(renderer, chart);
This will render distorted SWT charts because the run method makes unnecessary chart model modifications. Instead the developer should execute the following code:
	//Create an instance of the SWT renderer
	IRenderer renderer = new SWTGraphicViewer(parent, SWT.DIALOG_TRIM );

	//show the chart in the SWT renderer
	((SWTGraphicViewer)renderer).setContent((IDItem)chart);


Copyright © 2005 IBM Corporation and others.
All rights reserved. This program and the accompanying materials
are made available under the terms of the Eclipse Public License v1.0
which is available at http://www.eclipse.org/legal/epl-v10.html