On Dec 15, 5:57 pm, Bradbev <brad.beveri...@gmail.com> wrote:
> I have the following scenario:
> - a server that is listening on a socket for incoming connections.
> - when the server accepts a connection it uses send-off to run a
> handler function to handle the connection
> - the handler function loops using recur to handle packets
> - the handler function uses send-off for some packets so it can
> render a gui
>
> I know now that this won't work because when code running in an agent
> uses send or send-off, those sends are not dispatched until the agent
> changes state (ie, when the function returns). Because I am looping
> in my handler, none of its send-offs work.
>
> Would it be a good rule of thumb to say "agents should not use recur,
> but instead should use send-off to re-schedule themselves"?
For now, yes if they send other actions.
> Is this a performance impact? (ie, recur vs send-off)
Yes.
> Why are sends held until the agent state changes?
Because it is a common and otherwise impossible-to-implement case to
send actions to others after your state has changed (i.e. your new
state is the value of your final expression).
>Is it a technical reason, or a conceptual one?
It was done to facilitate the above scenario, but there isn't a
technical reason to preclude the other. If you look at the todo list:
http://richhickey.backpackit.com/pub/1597914
you'll see an entry for release-pending-sends, the API I'm considering
for enabling sends prior to state change.
Rich
--~--~---------~--~----~------------~-------~--~----~
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
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
-~----------~----~----~----~------~----~------~--~---
