Re: [boost] Re: Threads

From: David Abrahams (abrahams_at_[hidden])
Date: 2000-08-25 09:27:50

Maybe you and I have something different in mind. When I say "informal" I
don't mean "unrigorous" in any way. A formal correctness proof has been
shown to impracticable even for small programs without concurrency.

I am thinking, for example, about the power of the basic/strong/no-throw
exception-safety distinctions, and how with those three simple concepts you
can easily reason about the exception-handling behavior of large systems. I
don't know if there is an analogue in the domain of concurrent programming,
but I think it is worth searching for.

-Dave

----- Original Message -----
From: "Bill Rutiser" <wru_at_[hidden]>
> Let me add a warning about informal reasoning about concurrent
> processes. There exist clever, efficient, synchronization schemes,
> proved to be correct for two processes, that fail badly for three.
>
> -- Bill Rutiser