Update on 3rd Nov Go live

Share

Shipped the first end-to-end version of QA Monk yesterday.

2 private repos for now.

1 is for QA Monk with all the general guides.

The other is the implementation for a specific application.

For now no database, just a csv in a repo that I change with pull requests.

Not ideal but good for now.

So here is the high-level flow that got pushed.

1. Generate the test cases (Using Buildprint)

2. Run the cases (using Buildprint - uses browser-agent)

3. Mark them Pass/Fail

4. Once everything is done, create a deterministic Playwright test case from the browser-agent output.

This worked but has major issues.

Firstly, since we are using browser-agent, it pretty much has to do fresh trial and error using AI to create the Playwright code.

So I swapped that out in the next version.

Now we will use the Playwright CLI to do the initial walk.

Then compile the commands and try to re-run the code.

This seems to be working for now.

Factor-based combinations and permutations

Made a lot of other improvements, but the main one was this.

Mainly structuring tests by Factors.

Factors are main components that create branching flows.

Like user logged in, a specific payment provider.

The entire test case can be boiled down into these.

This concept is mainly from the following repo.

GitHub - microsoft/pict: Pairwise Independent Combinatorial Tool
Pairwise Independent Combinatorial Tool. Contribute to microsoft/pict development by creating an account on GitHub.

My version is just a bad implementation of it.