Thanks! The TLA+ modeling actually caught 3 bugs even before I wrote any code. Worth the upfront investment. It's way easier to debug a state machine model than distributed sync logic.
Happy to share more about the verification approach if you're interested!
VERY interested, are you planning on writing a blog? If so I would definitely read that post.
Please do in the repo, and thank you for the wonderful contribution on multiple fronts. This is very well done work and documentation. A few tips to help others discover the value on the Readme: Add the key memory benchmark (currently just above the acknowledgments) to the top of the Readme and link to the rest of the benchmarks on the page that you already have (maybe rename it benchmarks.md in the process): https://github.com/Dancode-188/synckit/blob/main/docs/guides...