On Sat, Apr 17, 2010 at 11:07 PM, alux <alu...@googlemail.com> wrote: > Hello Per, > > thats very cool! Many thanks! Completely solves my problem. (As far as > I can see :-) > > I'm not sure whether I completely understand the implications of the > nondeterminism you described; will think about it. You say "When each > promise can only be written by a single writer.." - I assume you mean > this as "Even if ..". Or am I on a wrong track?
No. The issue I was getting at with that statement is that when a promise is accessible to multiple potential writers there is a race as to who delivers on the promise first. All the runner-ups will fail. Who is the winner and who are the runner-ups is nondeterministic. It should probably be considered a bug in your code if that possibility even exists. In fact, I would argue for separating the capability to write a promise and a read promise, so that you can prevent this from happening accidentally. You can then hand out the reading capability liberally while being more careful with the writing capability. As for the nondeterminism issue with readiness checking: Take the simplest case of a single reader and a single writer. If the reader's only option is to block on the read then it works like Unix pipes: the relative schedule interleaving of the reader and writer is irrelevant to the ultimate outcome. However, if the reader can check to see if the promise is ready (the analogue in the Unix world would be if the reader does a select() on the file descriptor before reading it) then there is at least the potential for a harmful kind of nondeterminism where the outcome changes materially depending on whether the reader or writer wins the race. -Per -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en
