E2E Power Platform testing using GitHub and Playwright - advanced

Canvas Apps are particularly challenging for UI and End-to-End (E2E) test automation because Microsoft Power Apps renders much of the application UI within a Shadow DOM. While Shadow DOM provides benefits for Microsoft and app makers, it introduces several obstacles for traditional test automation tools.

Playwright has several key advantages over Selenium and other test tools, mainly around selectors and planning. The selector approach works much better against Canvas Apps that utilise a Shadow DOM.

Selenium tests were flaky as they rely on CSS and XPath to locate elements in the DOM.  Playwright allows users to navigate the DOM within the Page DOM, offering 6-7 ways to find an element, including getByRole().  Selenium often requires many waits/pauses to ensure the DOM is fully loaded, whereas Playwright uses the Page Object Model (POM), and I don't tend to need to use waits on a page.

Define Test Criteria using Gherkin - Go to GitHub Copilot

define gherkin test cases for the following feature: Check User is an administrator

// Generated response

Feature: Check a user can login and determine if they are an administrator

​ 

  Scenario: Check if the logged in user is an admin

    Given the system has started 

    When the user is logged in 

5    Wh 

e 

    Then the system should show if the current users is an admin or not

*Repeat for multiple features

I believe pairing LLM's to generate context-aware Playwright tests will allow developers and QA engineers to test apps better and faster.

GitHub Spaces  (Not too sure if I like GitHub Spaces)

"Put your code, docs, and notes where they belong: together. With GitHub Copilot Spaces, your AI pair programmer uses that context to deliver better, more tailored answers for you and your team".  Swap between models like Claude 3.7 Sonnet, OpenAI o1, and Google Gemini 2.0 

GitHub Repos, Actions, and Copilot work with VS Code, Visual Studio, and Eclipse.