Hi François, That makes sense. I think I had read that discussion a while ago but hadn't really grokked it.
Thanks! Juan On Tuesday, March 11, 2014 10:42:04 AM UTC-3, François Rey wrote: > > On 10/03/14 15:37, juan.facorro wrote: > > I have taken this approach as well, but I can can't seem to find a good > answer to this question: When you create a bunch of elements with their > corresponding channels to handle certain events, how do you handle the > closing of those channels and the termination of the related *go* blocks > once you remove the elements? > > > I'm not a specialist of core.async (so anyone please correct me if I'm > wrong) but here's what I understand. > > Go blocks aggregate into FSMs which can be garbage > collected<http://stackoverflow.com/questions/18800440/javascript-and-garbage-collection>like > any other objects. When parked go blocks are queued into the channel. > So what it comes down to is being mindful of where you keep references to > channels. More precisely, don't hold a reference to the channel beyond the > scope of its producers and its reading go blocks. That probably means > locality of readers (go blocks) is often preferable to a longer-lived go > block. I guess it's the same mindfulness that goes on when keeping > references to call-back handlers. > > See also this discussion: > https://groups.google.com/forum/#!topic/clojure/_KzEoq0XcHQ > > From Timothy you'll read that go block are garbage collected: > > When go's are parked, they are put into a queue on the channel, thus when > the channel is unreachable (besides inside the go block) both are > collected, and the thread of execution is effectively terminated. > > > From Brandon Bloom you'll read that: > > Querying the state of a channel at worst leads to race conditions and at > best leads to bad design. > You're only supposed to close a channel from the producer side. So if > you're the only writer, then you know if you've closed the channel or not. > If there are multiple writers, then need to be coordinated in some way. > Typically, they would alt! against reading from a control channel and > writing to the output channel. When you get a shutdown signal from the > control channel, you stop writing. > (...) > It is a programming error to write a message to a closed channel. "close" > is not a resource cleanup operation, it is a control signal. It "flows" in > the same direction as the messages sent on the channel itself. If the > receiver were allowed to close the channel, then the sender would have no > way of avoiding a closed/write race condition. > > > > -- 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 --- You received this message because you are subscribed to the Google Groups "Clojure" group. To unsubscribe from this group and stop receiving emails from it, send an email to clojure+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
