QF-Test 4.1.3 Released

Download-Link for QF-Test 4.1.3

Despite being just a "Minor Update", this release is a milestone for web testing with QF-Test in which our main focus was on performance. In WebDriver mode - for which recording has now stabilized - and for JxBrowser the improvements are enormous and they should also be noticable for other technologies.

Besides the usual bug fixes there are further interesting new features, including an improved variant of recording and executing double clicks.

Note: Selenium WebDriver for Firefox only supports the latest two Firefox versions plus the most recent ESR release. At the point of this QF-Test release that means Firefox 51, 52 and 52ESR. We try to improve on that as best we can and still support Firefox 38ESR, 45 and higher, but older Firefox versions simply don't run as well as current ones. There was a major change for Selenium WebDriver between Firefox 47 and 48 and WebDriver based tests run much faster with Firefox 48 upwards. For long-term support we currently recommend to use Firefox version 52ESR.

The detailed list of differences between QF-Test version 4.1.2 to 4.1.3 can be found in the Release Notes in section 3 of this newsletter or online.

Users with a valid QF-Test license of 4.1 can directly update to QF-Test version 4.1.3.

Customers with a valid maintenance agreement who have not yet upgraded their license to QF-Test 4.1 can request a license update via qftest@remove-this.qfs.de.

New Features:

• A major focus for this release is performance, especially for web testing. It has been drastically improved for WebDriver mode and JxBrowser and also significantly for web testing in general. The extent of some of the effects depends on the actual application and its implementation.
• Thanks to the performance improvements, recording in WebDriver mode has stabilized and should now produce the same results as in QF-Driver mode.
• Recording and replay of mouse double clicks is now more intuitive and consistent across all GUI engines. In the past, a 'Mouse event' node with event type 'Mouse click' and 'Click count' set to 2 was shown as a 'Double click' event, even though it was only the second of the two clicks required to simulate a double click. Such nodes are now called 'Followup click' and there is a new event type 'Double click' that properly simulates all the individual events for a complete double click in a single node.
• Test-suites can now be closed by clicking on their tab with the middle mouse button.
• The procedure qfs.database.executeSelectStatement in the QF-Test standard library qfs.qft has been improved with respect to the usage of a property group for the result set. The group is now cleared before being populated with the current result and the number of records is available via the property 'count'.
• It is now possible to configure fallback values for placeholders in the template file for recording procedures.
• QF-Test now shows the full path to the current test-suite as well as the QF-Test version in the title bar of the main window. This can be reduced to just the file name via the option "Show complete file path and QF-Test version in the title bar".
• WebDriver was updated for the latest Firefox and Chrome versions and QF-Test is now also compatible with the version of Microsoft Edge Browser shipped with the upcoming Windows 10 "Creators Update".

Bugs Fixed:

• Up to now resolvers were registered globally, applying to all GUI engines, which could lead to unwanted side-effects in hybrid applications. By setting the 'GUI engine' attribute in the 'SUT script' node that registers the resolver it is now possible to restrict the resolver to the desired engine.
• QF-Test now works around an issue in Jython 2.7 that prevented QF-Test startup in case the user's home directory contained international characters.
• Groovy was updated to version 2.4.8 which fixes a leak of PermGen space in Java 7.
• When stopping recording too quickly, the last recorded events, checks, components or procedures could get lost. Synchronization with the SUT has been improved so that this should now happen only in cases of extreme load.
• When switching between projects the expansion state of the tree nodes in the project view was lost.
• Summary output of a batch run with -verbose [<level>] did not conform to the information displayed in the status bar and in the report.
• JxBrowser components with preloaded content are properly handled now.
• 'Text input' nodes now also work for embedded browsers (FX WebView, JxBrowser and SWT embedded browsers) if 'Replay single events' is deactivated. Replay of single events also mostly works, but the events have to be simulated as hard events which is currently possible on Windows only.
• Some TEXT nodes in a web application were treated as disabled in WebDriver mode.
• After page modifications the WebDriver-API did not always assign the correct QF-Test component for a WebDriver element.
• Closing a secondary window of an Edge browser could inadvertently terminate the whole browser process. Also, browser shutdown in WebDriver mode has been improved for all browsers.
• Several issues for the embedded WebView browser in JavaFX have been fixed, including mutation handling, refresh and goto commands and return values from evalJS.
• Firefox settings are now only made persistent in the test profile if the system property "qftest.web.mozilla.save_profile" is set to true during browser startup.
• The QF-Test installation now includes libraries for the embedded Chrome browser in QF-Driver mode that are required for hardware accelerated graphics.
• On Linux, tests for Eclipse with embedded JavaFX components could cause a hard deadlock at Gtk level. This has been fixed for Eclipse/SWT versions 4.3 and higher.