2

I have a desktop application for which I need to automate the smoke test suite consisting of 30 odd test cases. Application consists of 3 screens with some buttons and data grid. Major work in this is to identify and verify the data in the data grids and clicking on some tabs and buttons.

I have done POC on following tools :

  1. Teststack.white
  2. Winnium
  3. TestComplete

First 2 tools are taking around 4 minutes to identify a single element in my application (couldn't find out why) TestComplete is perfect and is working seamlessly with my application, however, cost is a huge angle associated with it.

Recently, I got to know about Sikuli as well. Is it a good idea to create the smoke pack with sikuli, I think a lot of flakiness would be introduced due to this ? Also, any other tool that is successful in desktop automation?

0

I have used Test Stack White sometime back to automate a Windows based application. It was really fast, efficient and did the job for me.

I was able to interact with 20-30 elements in my application with less than a minute time. So I think the performance issue you are facing is specific to your application and maybe you can tackle it with a better programming approach or with the help of a developer.

White is also very extendable and easy to use if you have coding knowledge.

  • does it have to do anything with the kind of programming that dev has done ? its just interacting with the elements ? – brij Feb 23 at 12:15
0

I have had success in the past using AutoIT. Don't let the talk of their scripting language distract you though. It also comes with a DLL/COM to allow you to use virtually any high or low level code.

0

Recently, I got to know about Sikuli as well. Is it a good idea to create the smoke pack with sikuli, I think a lot of flakiness would be introduced due to this ?

I am using kantu with the desktop automation xmodule. Like Sikuli, it is image-driven and the automation runs stable and reliable. Also, I can move the tests between different machines - and they still run ok.

Kantu and Sikuli have the option to use text recognition (OCR). This is useful if your are looking for text that changes often, so you do not have to make an image of every option. Using text instead of images can improve the stability and the maintainability of a test significantly. It is also good for screen scraping (extract text from the app under test).

First 2 tools are taking around 4 minutes to identify a single element in my application

Kantu takes around ~1 second per image recognition step on your average PC screen. If I remember correctly, Sikuli has about the same recognition speed.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.