Wait the author says that this isn’t a bug in Paxos but one that he intentionally and arbitrarily introduced? The ‘bug’ here is maybe a bit of ambiguity from a book?
Edit: I appreciate being corrected on this.
Edit: I appreciate being corrected on this.
Emphasis mine, the bug is not in the algorithm but in the way it is described in the paper. I can see how the author's choice of words is a bit confusing here.
> So if you don't quite follow the letter of the text about acceptor selection, and then do follow the text about how acceptors handle accept messages, then you end up with the bug
Note the "don't quite"
Then again I don’t subscribe to the idea that doxumented misbehaviour is fine do ymmv.