re: deadlocks (Re: Wikipedia Merge Proposal)

[Raoul Duke]

On Thu, Apr 12, 2012 at 10:30 AM, Paul Morrison
<paul.m...@rogers.com> wrote:
> FBP deadlocks (as opposed to transaction deadlocks) are always a design
> problem.

ja, dataflow supposedly is much better than imperative when it comes
to deadlocks because the program runs the same way ever time.
(although i am not sure i believe that since presumably that isn't
true when you have e.g. human inputs.)

http://en.wikipedia.org/wiki/Oz_%28programming_language%29#Dataflow_variables_and_declarative_concurrency

[Paul Morrison]

Actually, Dan and Raoul, I see that in the 2nd edition of my book I
included several pages by Mike Beckerle on handling deadlocks with
_large_ amounts of data - "Parallelism Issues within Chap. 16 in the
current edition. I must confess I had forgotten about his work :-[