End-to-end correctness does not matter in a comparison. If every app implements the same functionality, then that is a valid comparison. Even if that functionality is incorrect.
If five apps implement an authentication flow, 4 relying on the framework and one building it from scratch because the framework doesn't provide one, then surely that should be factored into any comparisons.
If a framework comes with an opinionated templating system whereas another requires me to come up with a way to have nested templating (and whatever else I might require), then again, that should be a factor in my comparison.
A comparison of how easy or cumbersome it is to write production grade apps between frameworks is much more valuable.
Otherwise we are back to comparing TODO demos.