0

My problem is the following: We use Selenium IDE and RC at work to test the functionality of our website. Our website features a list that is saved on the server every time you add an item.It also allows users to log in.

When my selenium suite fails halfway (for example, because there was an error on the site), it may leave some items on the list, it may leave a user logged in, etc. The teardown scripts (log out, clear the list) are now part of the suite, but they wont be executed when the suite fails halfway. So when the next test starts, it does not start from scratch, as it should.This of course will make several steps in that suite fail, causing the whole test run to crash and burn.

One of the solutions I came up with was to, after each test suite, run several smaller teardown suites that are allowed to fail.

Example: Big Main Suite Fail Log out Fail clear list Success

This is what would happen on a test that fails halfway and left some items on the list

Big Main Suite Success Log out Fail clear list Fail

This is what would happen on a successful run.

Even though this solution would probably work, it's really terrible for your reports.

So I basically have 2 questions:

  1. What is the most elegant way of making sure that a suite starts from scratch, even though the suite before it failed?
  2. Is it possible to suppress "failure"messages of suites and cases that you don't mind failing?
Uwe Keim
  • 39,551
  • 56
  • 175
  • 291
PowerBI_Jaap
  • 43
  • 7

1 Answers1

1

For question 1. What is the most elegant way of making sure that a suite starts from scratch, even though the suite before it failed?

I would suggest you use testNG or JUnit framework. Both of these have annotations which will help you write what should happen whenever a test completes regardless of the test result. I have used only testNG. testNG provides additional listeners where you can define what should happen when a method passes, skips or fails.

You can also define methods which will be executed before any method or class is invoked (@BeforeMethod annotations).

I don't know how you are handling your reporting now, but testNG has a good HTML report which shows the basic details.

By using testNG you can achieve whats stated in question 2. testNG allows you to group your tests and run those tests together. You can add all the tests which you dont mind failing to one group and run it.

Learning curve is relatively less and it would make your job easier. You can find more details about testng and its integration with selenium here

A.J
  • 4,929
  • 2
  • 27
  • 36