IRC channel logs
2026-06-17.log
back to list of logs
<dthompson>doing pretty well! just getting back to work after a quick trip to toronto over the weekend <dthompson>nope, we were speaking at a toronto games week event on saturday <glenneth>Glad you're here actually dthompson. I was inspired to investigate FRP. I'm working on a web framework. Your talk at FOSSDEM gave me the nudge I needed <dthompson>still a lot of problems to figure out. maybe you can solve some of them :) <glenneth>I think I've got the diamond problem solved, still more work to do though <dthompson>one thing I've been kind of hung up on is propagation of timestamps vs. the actual values. <glenneth>I implemented a height-based topological scheduling. Every cell tracks its height in the dependency graph. Base cells are 0, computed cells are max(dependency heights) + 1 <dthompson>in alexey radul's thesis, he mentions that topological sort is how most FRP libraries work but it's rather complicated and restricts graphs to DAGs <glenneth>From what I have so far I do have basically glitch free propogation. I even wrote an example app using the framework, it's not being pushed hard but it works <glenneth>Really, I just wanted to say thank you. Inspirational talk you gave <dthompson>topological sort does get you glitch-free propagation <glenneth>wip, obviously lol. But I think it has legs <dthompson>if you have any breakthroughs let me know :) <glenneth>I called out your talk in a blog post for it :) Credit where it's due <glenneth>niche subject but incredibly interesting