One thing about this PR, since it was referenced in some recent bugs. This PR was supposed to just add the keyword "or", but it actually added a sizeable chunk of code to the parser. What was the reason for that? Can you explain why that was a necessity?

"arbitrary logical expression in guards." - see added tests.

In short, before this PR it was impossible to write something like when not((a) and (b)).
By now all regression bugs introduced with and after 2.6.x are fixed. The only current critical thing to concern of (if you're going to make a release) is just #2881 (I'm afraid I would not be able to install the latest node soon so I'm pretty useless for that part, sorry).

Thanks for the explanation. I did end up doing a release, but may have been premature. (See: #2896)