I need to double check this - I think it's the right fix, I think I may have overridden this method within my own code for that same reason.

Confirmed that yes, this is the right fix.

However, we don't have a test failing for this case, so we really should create that test case. @hudecof is that something that you could tackle this week or next?

I did not want to close this ;)
I could try, to write the test, but bit sure about the time line ;(

Running the GHCI to test correctness.

Until we create the test to ensure this won't happen again, I won't merge it though.

I hope I can find some time to write the necessary tests if you're occupied @hudecof

This still use the old GHCI definition file, resulting in very slow qa_docs (and usually ends up in failure)

If you can rebase this to the latest master, @hudecof , I can rerun the tests.

Though again, until tests are written to prevent same issue from happening, I won't merge this.

But at least we can be sure that the changes do not cause any regression.