News for hackers. Stuff that matters.
Re:TLA+: design, model, document, and verify concurrent systems by bayareanative on Friday May 03, 2019 @09:36PM (#19823721)
Interesting. I've worked concurrent systems in VHDL and Verilog (but not SystemVerilog) stacks, and those using C, C++, Haskell, OCaml, Ruby, Go, Erlang and Rust.

I wonder if this approach could formally verify systems like seL4.

Here's an approach to verifying concurrent systems using coq:

https://www.sciencedirect.com/science/article/pii/S157106610...

Re:TLA+: design, model, document, and verify concurrent systems by pron on Saturday May 04, 2019 @05:57AM (#19825485)
The main difference between Coq and TLA+ is that Coq is a research tool, intended and designed for use by researchers, while TLA+ is intended as an industry tool, and designed for use by engineers. They are essentially equivalent in their power when it comes to specifying and verifying software (although Coq is more suitable for proving general mathematics), but Coq is used almost exclusively by researchers. The learning curve is very different (one could get productive to the point they specify real nontrivial systems after learning TLA+ within a few days or a couple of weeks, while with Coq this takes many months), and TLA+ has much more automation.
Re:TLA+: design, model, document, and verify concurrent systems by hwayne on Friday May 03, 2019 @10:00PM (#19823822)
You might be interested in TLAPS, the TLA+ proof system: https://tla.msr-inria.inria.fr/tlaps/content/Home.html

There's also this essay on proving TLA+ specifications in Isabelle: https://davecturner.github.io/2018/02/12/tla-in-isabelle.htm...