Re: Concurrency always is non-deterministic?

[Phil Clayton <phil.clayton@lineone.net>]

On Feb 13, 5:41 pm, Long Hoàng Đình <long....@gmail.com> wrote:
> The author of this post said so.

The author's point is that non-determinism introduced by concurrency
requires synchronization to make the overall program deterministic but
this synchronization is difficult to get right (so why not use the
Glasgow Haskell Compiler that takes care of all that for you...)


> If Ada were non-deterministic at concurrency, I guess it couldn't be real-time, right?

If you wrote a concurrent program but left out the synchronization
required to ensure deterministic behaviour, in general, the result
would be so unpredictable that it doesn't make sense to ask whether it
is still real-time.

Synchronization support exists for many languages, e.g. via the
pthreads library.  In Ada, tasking (and its synchronization support)
is actually part of the language which has numerous advantages.


> http://ghcmutterings.wordpress.com/2009/10/06/parallelism-concurrency/
>
> Please let me know your opinions about that post.

I am a big proponent of functional programming and it does have
advantages for parallelism.  I find the discussion on parallelism vs
concurrency nonsensical.  I'm sure GHC will have some great new
parallelism features though.  The post should be seen in a certain
software context: functional programming is unsuitable for systems
that are any of: real-time, embedded or critical (I will happily
expand on that claim) so any claims made by the author only extend so
far.

Phil