[dmd-concurrency] What is protected by synchronization?

[Michel Fortin]

Le 2010-01-31 à 11:39, Robert Jacques a écrit :

> Third, I would argue that a some-what safe system is just as unusable, if not more so, than a slow solution. Until you can legitimately take down the 'here be dragons' sign, you'll have all the problems of threading today. And even PhDs can't get shared-state threading right today. Which is why every single article/books/blog on concurrency says to avoid shared-state like the plague.

Actually that's a very strong argument against having synchronization as a language primitive. Perhaps indeed if we can't make the synchronization primitives powerful enough to support the typical use case decently while keeping everything safe we should just remove synchronization from the language.


-- 
Michel Fortin
michel.fortin at michelf.com
http://michelf.com/