Selenium Testing your Kubernetes Apps with Machine Learning and Testim

Editor's Notes

1. 1) Talk about the challenges with UI testing - Oren 2) Talk about how machine learning can solve them - Oren 3) Introduce test application - Dan 4) Introduce Testim and how it works, create some tests. - Oren 5) Introduce Codefresh, add testing step to use Testim - Dan 6) Q/A
2. Thank you everyone for inviting me. My name is Oren and I’m the founder and CEO of Testim. I’ve been building products for the last 20+ years, focusing on developer tools. And I have to say that the testing challenge is the one of the hardest I’ve encountered, and that why I’m excited being here
3. Shift left slide 1/3: Dan: Oren, can you start by explaining to the audience what is Shift Left and why is it starting to gain momentum recently? Oren: Shift left is the name of a development paradigm which encourages finding issues as early as possible. The initial premise is that the earlier you discover a bug, the less effort & cost you need to fix it. This ranges from a requirement fix which might take a few hours and sometimes minutes, while before release it will take X10 or X50. In production it’s a total different case. I’m not even mentioning hardware, where a production mistake costs billions or human lifes When you have everything set up, I mean testing early, and have everything automated (including the deployment), you actually get to what we call continuous deployment Dan: And what’s the adoption rate in the software industry?
4. Shift left 2/3 Oren: We see the leading companies, such as Google and Facebook being already there, but most companies are struggling to get there (although they’re trying). Let me show the same survey result in a different way
5. shift left 3/3 Oren: you can see the majority are not continuous deployment Dan: what are the current challenges which hinder companies?
6. Challenges with UI Testing (first 2 min our of 10 min)Oren: there are a lot of surveys that show several challenges, and I’ve work in the recent years to map them onto the different stages of a company
7. Challenges with UI Testing (first 2 min our of 10 min)Oren: there are a lot of surveys that show several challenges, and I’ve work in the recent years to map them onto the different stages of a companyOren:the laggers find trouble even finding good hiresThen people struggling with authoring the tests, as it’s mostly asynchronous, as a human you don’t just click on stuff, you (unconsciously) wait for something to appearInitialization is the next challenge, you created a test which adds goes to ebay and adds something to the cart, and you validate that there’s one item. It will fail on the next run if you don’t reset that (because they will be two on the cart)Next is the maintenance, E2E and UI testing are considerate fragile, because on every code change the tests might break. Dan: so how does AI and machine learning help with that?
8. Dan (?): and this is where the AI comes in I guess, doing things that humans can, like looking at multiple attributesOren: exactly, a machine can look at all the different attributes, which could be hundreds. This means that when a class/id/text/some-attribute changes, it doesn’t fail the test. A machine today can learn from each execution, and know that a specific attribute is significant or now. For example. Everyone used to say that “ids” are the best property to use in order to find elements. A machine can tell you whether it’s stable over the last month, or is it randomly generated on every build
9. Introduce test app (5 min) Dan: When microservices came on the scene it actually made shift left more complex because we needed a way to pull in our full application that was now split with lots of microservices. For this Demo we’ll use a Voting App packaged as a Helm Chart. If you’re not familiar with Helm...
10. Automated Demo (10 min)
11. Automation (introduce demo and Codefresh) (5 min) Automated Demo (10 min)
12. Automation (introduce demo and Codefresh) (5 min) Automated Demo (10 min)