Known issues - Run sessions

Relevant for: GUI and API tests

Local UFT One run sessions

Multiple instances of UFT One on the same machine UFT One does not support running tests from multiple instances of UFT One simultaneously on the same machine, either in the same Windows session or in different Windows sessions.
Local machine

When running UFT One tests or components on a local machine, if the computer on which the application is being tested is locked, your test run may fail.

Workaround: Install UFT One on a virtual machine (without a screen saver or lock password), and start or schedule your run session on the virtual machine. Then you can lock your local computer without locking the virtual machine.

Test Batch Runner / UAC

It is not recommended to use Test Batch Runner with the UAC (User Account Control) feature set to ON.

Back to top

Remote UFT One run sessions

Citrix XenApp support

If you are testing applications launched by Citrix XenApp, UFT One must also be launched by Citrix XenApp.

Logged off RDC / Citrix

If you are running UFT One on a remote machine using Remote Desktop Connection (RDC) or Citrix, we recommend not logging off your remote session.

Logging off your remote session while UFT One is running may cause the following issues to occur:

  • The test or component run session may fail.

  • Steps that contain keyboard or focus operations may fail.

  • The still image capture and/or the Screen Recorder (in the run results) may display a black screen.

  • Steps for which the device level replay is configured to use the mouse (instead of browser events) to run mouse operations may fail.

Configuring credentials for the remote session that are not the logged-in user's credentials

If one user starts a remote UFT One run session, and then the session locks or disconnects, the run continues using the credentials defined in the UFT One Options (Tools > Options > General tab > Run Sessions).

If these credentials are associated with fewer permissions than the user who started the session, the test run may encounter permission issues.

Keeping this in mind, you can still configure credentials other than the ones you used to log in to the UFT One computer, to be used for these remote sessions.

Multi-use remote credentials

If you are running UFT One and your application on a remote computer that supports multiple users logging in with the same username and password, we recommend not disconnecting from the remote computer.

In such cases, disconnecting and reconnecting will open a new user session, and your UFT One run session may be lost.

Keep your connection open to retain your UFT One session and run results.

Windows operating system issue

If you encounter issues when running UFT One remotely, modify the settings in the Group Policy Editor (GPE) Microsoft Management Console (MMC) snap-in as follows:

  1. On the UFT One computer, go to: Computer Configuration > Administrative TemplatesWindows Components > Windows Logon Options > Disable or enable software Secure Attention Sequence

  2. Enable software Secure Attention Sequence.

  3. Select the Service and Ease of Access applications option.

  4. Restart the machine.

Caution: Enabling this configuration can present a security risk as it allows applications and servers to login to the UFT One computer remotely. Make sure your UFT One machine does not contain or provide access to sensitive information.

ALM Test Lab

If you are running GUI tests from the ALM Test Lab, you must select the Allow connections from computers running any version of Remote Desktop option in the Windows Remote Settings (Control Panel > System > Remote Settings)

Authentication errors

If your test run fails because of authentication errors:

  • Restart the UFT One Helper Service in Local Services

  • Disable the registry keys under [HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\System]

    "legalnoticecaption"="<anytext>"

    "legalnoticetext"=”<anytext>"

Back to top

Running API Tests

Calling a GUI test

If you are running an API test that is stored in ALM and calls a GUI test, you must either ensure that UFT One is open with a solution open, or you must close UFT One and allow ALM to open it.

The test will not run if UFT is open with no solution.

Calling an API test

If your API test is called by a GUI test, values received from the API test output parameters are not supported as values for variables defined in functions in your GUI test.

External .dll files

When you manually add a reference to an external .dll, UFT One prompts you to save it locally. To change your preference about a specific referenced file, remove the reference and add it again manually.

Shared folders

Running tests on remote machines using shared folders, may require adjusting the .NET 2.0 security settings.

Suggestion: Open the Control Panel and locate the Administrative Tools, either by browsing or through a search. In the list of Administrative Tools, look for the following entry: Microsoft .NET Framework 2.0 Configuration. If it is not present, you must install the .NET Framework 2.0 SDK.

Validate Structure checkpoint

The Validate Structure checkpoint fails if the expected value is a SOAP Fault and the Web Service call returns an UnsupportedMediaType status.

Load Test mode

When running an API test in Load Test mode, checkpoint verification in API test steps is not supported.

Updated log4net.dll

If your API tests were created and built in earlier UFT One versions, to continue to use your tests, you must upgrade them by opening them in a new version of UFT One.

Back to top

Running steps that contain Insight test objects

Requirement: Active session and application

The computer session must be active, and the application visible (and not minimized). for the steps to run successfully. This is because Insight uses data from the screen to compare to the images stored with your test.

Remote connection

If you are testing an application running on a remote computer, and you use a minimized Remote Desktop Connection window, these steps will fail.

Workaround: Use a different program (for example, Virtual Network Computing) instead of Remote Desktop Connection to run steps on a remote computer.

Display size / zoom levels

For Insight object identification to succeed, the display size defined in the operating system and the browser zoom levels (if working on a browser) must be the same when the test runs as they were when the objects were learned or recorded.

Excluded areas

When using excluded areas in Insight, the included area must contain enough significant content to enable object recognition.

If the remaining content is not detailed enough, UFT One may locate too few or too many matching controls on the screen.

Workaround: Do one or both of the following:

Parent Browser objects

When searching for an Insight object within a parent Browser test object, UFT One searches within the selected browser tab, not in the whole browser window.

Dual monitors

Insight is supported only on the primary monitor. Therefore, if you are working with dual monitors, make sure that your application is visible on the primary monitor when you use UFT for Insight test objects.

Back to top

Running steps that use OCR

The following limitations affect steps that are based on OCR, such as Text or Text area checkpoints, Text test objects, and methods that retrieve text from the application.

OCR configuration and optimization

Make sure that UFT One is configured for optimal text recognition. See:

Requirement: Active session and application

The computer session must be active, and the application visible (and not minimized) for the steps to run successfully. This is because OCR uses data from the screen to compare to the text stored with your test.

Remote connection

If you are testing an application running on a remote computer, and you use a minimized Remote Desktop Connection window, steps with Text test objects will fail.

Workaround: Use a different program (for example, Virtual Network Computing) instead of Remote Desktop Connection to run steps on a remote computer.

Performance
  • Running steps that use OCR takes longer than running steps that use property-based object recognition.
  • When running tests with Text test objects, we recommend turning off Windows' animations to improve performance.
Inaccurate text recognition
  • Some icons and elements will be recognized as special characters and their recognition may vary from one run to another.
  • When performing operations such as highlighting on recognized text, the operation location may be inaccurate in the following cases: 

    • There is a large space before or within the text.
    • In the font being used, there are big differences between the character widths.
.GetVisibleText test steps

testobject.GetVisibleText steps may not work properly on text objects enclosed in frames.

Workaround: Insert the following test step before the GetVisibleText step into your test to set the fieldMarkingType ABBYY parameter. This instructs the ABBYY OCR engine to extract texts from frames.

TextUtil.SetABBYYParameters "fieldMarkingType", True, 2

Back to top

Running steps that use AI Identification

The following known issue exists when running tests that contain AI-based steps:

AIUtil("search").Search method When running an AIUtil("search").Search method, if a search control in your application opens a input box that is not immediately ready to receive text input, some of the characters from the search string may be lost.
AI Transformation Assistant
  • When running on mobile applications, the AI Transformation Assistant does not scroll to look for objects. Therefore, when running a test with the AI Transformation Assistant, AI object suggestions are not provided for objects that are not visible on the current mobile app screen.

  • The AI Transformation Assistant is not working when you run tests on Chrome, Chromium Edge, and Firefox browsers.

    Workaround: Disable the Accelerate web testing on Chrome, Chromium Edge, and Firefox option in Tools > Options > GUI Testing > Web tab.

AI-based test script AI-based test scripts fail to run on Firefox browser if there is a status bar at the bottom of the browser.

Back to top

Running tests in parallel

The following known issues exist when running tests in parallel using the ParallelRunner tool.

Unsupported functionality

Edge

The ParallelRunner tool can run on only one Edge browser at a time. Therefore, if you have multiple tests running simultaneously, only one of them can be running on Edge.

Web 2.0 toolkits

All Web 2.0 toolkits (ASP .NET AJAX, Dojo, Ext-JS, GWT, jQueryUI, SiebelOpenUI, and YahooUI), are not supported when using ParallelRunner.

Automation object model

UFT One's automation object model is not supported for tests run using ParallelRunner.

AI-based web testing

AI-based testing on web applications is not supported for tests run using ParallelRunner.

Specific test objects and methods

.object method

When running a web script using UFT One's ParallelRunner, the .object method is not supported when Internet Explorer is running in a documentMode below 8.

WebFile objects

WebFile objects are not supported for parallel execution.

SystemUtil.Run methods

Browsers launched by the SystemUtil.Run method may have unexpected results when running tests in parallel.

We recommend using the WebUtil.LaunchBrowser method instead.

Run results

Internet Explorer

To view the ParallelRunner run results in Internet Explorer, use an Internet Explorer version 10 and higher.

Safari

Run results with device monitoring graphs do not show red lines indicating the time stamp, when opened in Safari on a Mac machine. To view these red lines, use a different browser to view the run results.

Device system monitoring

Device system monitoring is not displayed in HTML run results graphs. Instead, to view system monitoring data for device tests run in parallel, check the log file linked from the run results.

Application snapshot

In parallel testing on Java applications, ParallelRunner may save incorrect application snapshots to the run results report.

Device testing

Device reservations and release

  • ParallelRunner cannot reserve devices based on tests that are data-driven using the Device test object, or Util methods such as MobileUtil.SetActiveMobileDeviceByID.
  • During parallel testing, pressing CTRL+C does not release the device.

Insight / OCR / highlight

Parallel testing on mobile devices may fail when the test uses Insight objects, or employs the OCR or highlight mechanisms.

Workaround: For Insight objects or the OCR mechanism, arrange the remote access windows displaying the devices in parallel so that they do not overlap.

Troubleshooting

In some cases, web tests stop responding whenever they are run using ParallelRunner.

Workaround: 

Change the address that ParallelRunner uses as your local IP address.

  1. Obtain your computer's local IPV4 address (type ipconfig in a command window).
  2. In the <UFT One installation>\bin folder, open parallel.ini (or parallel.custom.ini if it exists).
  3. Change the IP=127.0.0.1 definition to the local IPV4 address and save the file.

Back to top

Running tests in Maintenance Mode

Using programmatic descriptions

When running in Maintenance Mode, if a step contains a programmatic description for an object that is not found in an application, it may take a while for Maintenance Mode to indicate there is a problem. If you use the option to point to the object, it may take a while for Maintenance Mode to reopen afterward.

Objects without mandatory or assistive properties

When the Maintenance Run Wizard cannot find an object, and the test object description for that object does not have any mandatory or assistive properties (it is identified only by its ordinal identifier, such as a Browser test object), then when you point to the object, the wizard is unable to fix the problem and displays a message that the object you pointed to has a test object description that is similar to the object that UFT One could not identify.

Workaround: Use the Update from Application option in the Object Repository window (for objects in the local respiratory) or in the Object Repository Manager (for objects in a shared object repository) to fix the test object description.

Updating a step by pointing to a different object class

When the Maintenance Run Wizard cannot find an object in the application, and you point to a different object class to replace it, the Maintenance Run Wizard offers to add a step with that object and the object's default method. However, the wizard does not insert any method arguments for the step. If the step method has required arguments and you accept the step that the Maintenance Run Wizard proposes without modifying it, the step fails when you run it.

Workaround: Enter valid method arguments for the step.

Updating XPath or CSS properties

When running in Maintenance Mode, UFT One may replace test objects with XPath or CSS identifier property values with new objects from your application.

Workaround: Use the Update from Application option in the Object Repository Manager to update specific test objects with XPath or CSS identifier property values.

Objects used in function library Maintenance Mode cannot fix problems in an object's properties, if the object is used in function library that is called by a step. If Maintenance Mode detects a problem in an object in a function library, a message is displayed indicating that the test is read-only and that Maintenance Mode is disabled.

Back to top

Running tests in Update Run mode

Insight

The Update Run mode does not update Insight or Text test objects.

To update an Insight or Text test object, open the object in the object repository, and click the Change Test Object Image button in the Test object image area.

For a Text test object, you can manually update the text property, which is used to identify the text in the application. Note that this does not update the image saved with the test object.

Checkpoints and output values

Update Run does not support updating complex checkpoint and output value types such as File and XML.

During the Update Run, these checkpoints and output values run as they would in a regular run session and will fail if there are differences between expected and actual values.

Back to top

Recovery scenarios

Relevant for: GUI tests and components

To run a recovery scenario in response to an application crash, the following conditions must be met:

  • The application that crashes must have a window and must be a 32-bit application.
  • Use a Recovery.Activate step in your test to explicitly check and perform the recovery scenario after the application crashes. For details, see Recovery Object in the Utility Objects section of the UFT One Object Model Reference for GUI Testing.

  • The Recovery Operation is Keyboard or mouse operation or Close application process.

Back to top