Thinking About Software Transactional Memory
So, Audrey’s gone and added STM to pugs. I’ve written before that Software Transactional Memory is the future of concurrent programming (contra explicit locking). In that same post, I plugged Rob Ennals’ paper about efficient STM. Reading that paper again today, I noticed that the design is aimed towards object-oriented systems. Well then.
Rubyists who know me know what I’m contemplating. I’m not sure exactly how it’ll work yet, but one thing’s already for certain: it’ll involve reified monads.