Full Workshop (2 days)

on YouTube ↗07h 15m

" summary

A full type-level chess implementation demonstrates what Haskell’s types can enforce, and how fragile GHC’s performance can be: a naive check for “king in check” makes an 8-move game compile in 1m37s using 23GB vs 24s and 8GB without it; adding move-number details in error messages blows a 19s/4.5GB compile to ~2m and 26GB; using type-level chess model + CPS EDSL plus a ray-based check replaces huge move lists, halves memory and cuts compile-time by up to 5x on some games while revealing unpredictable GHC behavior and a few reproducible pitfalls.

AI-written summary · about

" chapters (32)

1. 0:00Start
2. 3:29Toby Bailey
3. 15:01Algorithmic Improvements
4. 19:49Descriptive Error Messages
5. 20:29Custom Type Errors
6. 21:54Chess School Shorthand
7. 22:10Creating an Optional Shorthand Syntax
8. 28:21Conclusion
9. 30:14Jeremy Gibbons
10. 36:19Express Applications of Dynamically Typed Haskell Expressions
11. 40:10Evaluate
12. 40:34Heterogeneous Lists
13. 42:16Encoding Variables
14. 50:34Property-Based Testing
15. 54:31Synthesizing Functions That Match a Set of Argument Result Pairs
16. 1:08:42Automatic Function Inversion in Haskell
17. 1:09:39Function Inversion
18. 1:10:55Functional Patterns
19. 1:11:45How Is Inverse Computation Achieved
20. 1:13:15Pattern Match Compilation
21. 1:30:13Limitations to the Template Haskell Implementation
22. 1:34:19Is the Transformation Done on the Service Language or on Core
23. 1:35:10Why Do You Use the Fm Monitor
24. 1:49:14Retrofitting Linear Types to Haskell
25. 1:54:32Server Response Times
26. 1:54:47Latency Spikes
27. 1:56:48Memory Management Can Be Expensive
28. 2:00:00Bug Fix
29. 2:07:02Treat Sockets as Stateful Objects
30. 2:11:08The Virtues of Modularity
31. 2:11:52High Performance Language Interop
32. 2:12:17High Performance Language Interop