I’m a long time Selenium WebDriver.js user and been looking into the current state automated test frameworks.
What I’ve learnt:
The big divide is between those that use the
WebDriver classic system (WDIO, Selenium WebDriver.js etc) and those based on Websockets or Chrome Developer Tools (Puppeter, Playwright)
Playwright’s current issue is that it has to patch Firefox and Safari with the inherent problems that approach introduces.
Can we be sure that the development of Firefox and Safari will move to a point where that is no longer necessary before a competitor achieves the same integration ?
Mass adoption as a determiner of decisions is a doubled edged sword.
If it provides investment to improve the tooling and thoughtful adoption: yay!
But if it doesn’t we can also end in large sunk cost fallacy that eventually triggers a re-think and (as you rightly identify) calls for GOLD.
I like to try and base these decisions on an understanding of the technology and where it currently sits in the wider eco-system.
My humble opinion: Playwright is the winner
Reason: People support.
I’ve seen this before, it is not a valid reason to decide but it will do.
In the past the same happened many times, for example: Apache Mesos vs Kubernetes. Mesos was light-years ahead but people choosed K8s. Today nobody remember Mesos I guess
The Javascript community has historically been notably more fickle relationships with frameworks as those who’ve seen the rise and fall of jQuery, CoffeeScript, Angular, et al can attest
This might be a discussion around web test automation as well. So far we’re not really discussing desktop UI automation here, nor mobile app UI, nor game console testing.
On the desktop UI side, sad to see that MS seems to have abandoned their WinAppDriver, and we have multiple alternatives from the community instead. FlaUIWebDriver might be the one replacing the MS one over time but we’ll see.
And what about AI, isn’t AI based automation the talk of the future these days?
I know a couple of orgs that have moved either to Playwright or Cypress - they either find WebDriver too flaky and give up, or people want something new on their CV. The reality of WebDriver is that it rewards time and experience. Once you have strategies to handle all the exceptions you encounter, it’s actually pretty robust.
I do agree testing underlying APIs is more useful than testing at the UI. In my case, if only we had an underlying API - the front looks state of the art; what lies beneath is a little older (and refuses to die).
Webdriver has inherent issues with latency dues to its implementation on http request and response, with each command requiring a new http connection. This is only compounded when you use a remote service like BrowserStack or SauceLabs.
CDP/BIDI use of web sockets that remain open is going to reduce flakiness.
From my perspective, the future of automated testing in India feels like it’s on the cusp of something transformative, driven by AI, cloud tech, and the rise of low-code platforms. I’ve noticed how AI-powered tools, like Mabl and Testim, are gaining ground, especially as companies here lean into machine learning to automate testing more intelligently with self-healing scripts. Given India’s huge presence in IT services, I think tools like Jenkins and GitLab CI, which integrate well with CI/CD pipelines, will continue to be vital. Plus, it’s pretty exciting to see Indian-born platforms like BrowserStack leading the way in cloud-based testing, offering scalable and remote solutions that fit perfectly in this fast-paced environment. What I find especially interesting is how codeless tools, like Katalon and TestProject, are making automation more approachable, even for non-tech folks. It really aligns with India’s ongoing push toward upskilling and embracing digitalization at every level.
