Some of us must have been on that one project, where your test suite was causing more problem than solving it. You change one thing, and hundred tests will fail. Your continuous integration build will fail randomly, but will pass if you just re-trigger it. This eventually leads to people losing all the trust on the test suite. They stop adding tests, because it’s more painful than writing production code. They start ignoring failing tests, because they fail randomly and nobody knows why. Everybody knows tests are now more trouble than help. In this talk, we will talk about some behaviours and reasons which lead to this "flaky test suite" situation. What are some development practices, which can avoid such situation. And finally we will also talk about how to fix this situation if you are already in it.


Outline/Structure of the Talk

The flow of the session will be as follows :

1) Introduction

2) Why do we write tests?

3) What are non-deterministic tests?

4) Problems of non-deterministic tests

5) Detailed causes of non-determinism and solutions for solving them

6) Strategy to resolve the existing build flakiness on your team

7) Future scope (Poka yoke - Mistake proofing ) and conclusion

The entire session will be splurged with code examples and personal bad / good experiences related to test non-determinism.

Learning Outcome

1. Understand why it is important to write good test code, similar to writing good production code.

2. Understand and appreciate the problems which can lead to flakiness of tests and how to stay away from them.

3. Understand how to strategise and resolve the issue of flaky tests on your team if you are already plagued with the disease of "flakiness".

Target Audience

Developers, Quality Analysts, Managers

schedule Submitted 4 years ago

Public Feedback

comment Suggest improvements to the Speaker
  • Srinath Chandrasekharan
    By Srinath Chandrasekharan  ~  4 years ago
    reply Reply

    Hi Yahya,

    I read through your propsal and like the idea of focussing on tests. However I feel that a lot of this is theory and from the books. I believe that the audience at AI2016 will know about this and what would be interesting is to see how experience on the ground from projects helped to drive agility in the project.

    Hence I feel that there should be more examples from your experience . What are your thoughts on this and if  you agree, please update the proposal with the relevant data.




    • Yahya Poonawala
      By Yahya Poonawala  ~  4 years ago
      reply Reply

      Hi Srinath,

      Thanks for your comments. I do feel that each of the causes and solutions as well as the strategy is something we have seen and learned from our projects itself. So, the entire talk would be based on our experience from different situations and not just theory from books.

      However, if you feel that the audience might be knowing the stuff already would it make sense if I make the talk level "Beginner?".

  • Liked Isha Tripathi

    Isha Tripathi - Are we discriminating against men?

    Isha Tripathi
    Isha Tripathi
    schedule 4 years ago
    Sold Out!
    90 Mins

    The IT industry is evolving, and has been evolving for a while now. The one thing that has, however, not changed as drastically as one would expect within the industry are the diversity ratios. The software delivery stratosphere is still largely dominated by men. More and more organisations are now waking up to the reality of this situation, and are taking measures to fix diversity imbalance in their respective workspaces. One recurring argument against such measures to fix gender inclusivity is that these steps are discriminative against men. This, in my opinion, stems mostly from a lack of awareness about the challenges that women in the IT industry face on a daily basis. As part of this particular talk, I would like to bring to light not only some of these challenges, but also various myths about the reasons why there are fewer women in the industry and why this is a problem that we need to fix starting now.