Building a production-grade parser is an exercise in discipline. You need to translate thousands of grammar rules exactly. You need to catch subtle bugs that only surface on edge cases you've never seen. You need to verify every decision against a reference implementation. There are no shortcuts.

I know this because I've done it before. I led the effort to build the MySQL parser for Vitess. That took over a year with help from talented contributors. So when we needed a Postgres parser for Multigres, I expected a similar timeline.

It took eight weeks. 287,786 lines of code. 304 files. 130 commits. 71.2% test coverage. 2.5x faster than the cgo (Go's C interop) alternative.

We covered a lot of ground in this series. We started with the following objectives:

• Propose an alternate and more approachable definition of consensus.
• Expand the definition into concrete requirements.
• Break the problem down into goal-oriented rules.
• Provide algorithms and approaches to satisfy the rules with adequate explanations to prove correctness and safety.
• Show that existing algorithms are special cases of this generalization.

Below is a summary of the topics we covered.

This section covers topics that were deferred in the previous parts. However, these are necessary to deploy consensus systems in production.

This section discusses various approaches to achieving consistent reads in a consensus protocol.

This section discusses ways to change the durability rules for a consensus system. As you will see below, any approach that respects the governing rules is sufficient to safely effect these changes.

This section covers the second part of a leadership change: The discovery of the best timeline and its propagation to prepare for the next leadership.

This section covers ways to satisfy the prerequisites for a leader change: Revoke previous leaderships and recruit nodes for the new leader candidate.

In this section we will discuss the challenges of ordering decisions when multiple agents are involved.

In this section we will cover how to fulfill requests in a generalized consensus system.

This section discusses the fundamental governing rules of any consensus system.