QF-Test at Swiss Life

Since 2012, we utilize QF-Test for daily GUI regression testing of our new insurance quote system EVApro and are very satisfied.

For us, the use of QF-Test mainly has the following advantages:

  • Generic approach: With QF-Test it is possible to record actions elegantly for each occurring control type and then to abstract these, e.g. the actions 'set value' and 'check value' for text fields and combo boxes or the action 'click' for buttons.
  • Shared test case files for HTML and Java interfaces: The application under test exists in both an RAP and an RCP variant. Due to the generic approach it is possible to use the same test case files (*.xls) for the test of the two variants.
  • Simple recording within the application: Due to the generic approach the test case files contain only relatively simple data such as (set value, PersonDialog.BirthDate: DATE FIELD, 06/05/1978), through which a control with the specified ID obtains the specified date value. Such information is recorded automatically when a user operates the application; when he opens a special dialog of the application, a table will appear from which he can copy the information directly into his test case file.
  • Small, well structured, easily readable test suite file: We need for RAP and RCP only a single shared test suite file (*.qft), which is very small (approximately 150 KB) and well structured (XML). In addition, it is fully readable: In contrary to some QF-Test competitive products this file has no encrypted or proprietary sections. This also greatly facilitates the comparison of different development states of this file in our version management system.
  • Easy upgrade from RAP 1 to RAP 2: Due to the generic approach the switch from RAP 1 to RAP 2 was easy to accomplish: Only the test suite file had to be revised, and for most control types it was even sufficient to change just the component reference 'class', i.e. type of component, of the appropriate generic component.
  • Competent support: In several one-day workshops at our site an employee of Quality First Software GmbH (QFS) ably supported us to continuously develop the test suite file. Especially in case of more complex controls (such as tables), he helped us through profound product knowledge, in particular through the use of scripts. E-mail queries between the workshops were always quickly and competently answered by QFS.
  • Excellent documentation: Manual and Tutorial have often helped us. It is obvious that the authors do not view documentation as a chore, but are convinced of their product and want to describe it fully and understandably. They have succeeded there.
  • Very comfortable call options: Since variable values can be passed when calling QF-Test we are able to determine very comfortably via the parameters of a self-written batch file,
    • what test cases are meant to be run (all 400 test cases of all 35 fares of our insurance quote system? Only the test cases of a given tariff? Only a particular test case?),
    • which variant is to be tested (RAP or RCP ?)
    • what level of development is to be tested (released version or local unstable?)
    • whether to move the mouse during testing
    • whether QF-Test is to be started interactively or in the background
    • which browser to use in the version in the RAP case (Internet Explorer or Firefox?)
    • which QF-Test version to use.
    Each call parameter is either a (tariff or test case) number or a single letter, thus very simple. This way our department can easily create and run test cases without having to know intricate details.
  • Simple workarounds for version changes: For GUI regression tests several applications have to work together:
    • in the RAP variant: the browser in which the actual application is running (by default Internet Explorer)
    • the program in which format the test case files are present (Excel)
    • the GUI test tool (QF-Test).
    In these interactions problems can occur, of course, if the version of one of the involved application is changed. As a workaround, we therefore have, for example,
    • tested the RAP variant with a different version of QF-Test than the RCP variant
    • used instead of the current Internet Explorer version an older Firefox version in the RAP variant.
    Such workarounds so far always helped us to bridge the time until the test suite file was revised in a subsequent workshop, a new QF-Test version was provided or an improved Internet Explorer version was available. Thanks to the very comfortable methods mentioned above we could implement and undo these workarounds always relatively easy.

We can therefore absolutely recommend the usage of QF-Test.

Author: 

Dirk Stanke

Contact:
Max Gest
Application manager EVApro
Swiss Life AG, German Branch
Berliner Straße 85
80805 Munich

(Original German texts and citations are translated into English.)