Siri et. al.,
Thank you, I'm glad there is agreement that the functionality is desired.
In a similiar vein I'd like to see an enhancement to allow passing
information back in the exit message of a child. Currently a child's
init/1 function may return {ok,Child} or {ok,Child,Info} and
supervisor:start_child/2 returns the same. I'd like to see the inverse
where the child may exit with either 'shutdown' or {shutdown, Info},
in response to a shutdown request from it's supervisor, and 'normal'
or {normal, Info} for a normal exit.
The child may exit with any value now of course however if the exit
value isn't expected the supervisor restart strategy and SASL error
logging will not work as desired.
You may look to my radierl(*) project for an example of a practical
application of this functionality. In this stack application the
radius_server module implements a gen_server which will dynamically
start a new gen_fsm process, under supervision, to service requests.
It keeps the gen_fsm's pid in state stored in a gb_tree() with the
client's info as the key. In this way client requests can be quickly
looked up and forwarded to the gen_fsm. When the gen_fsm terminates
the gen_server needs to reap the state.
I implemented a clause in radius_server to efficiently do this:
handle_info({'EXIT', _Pid, {shutdown, Key}},
#state{handlers = Handlers} = State) ->
NewHandlers = gb_trees:delete(Key, Handlers),
NewState = State#state{handlers = NewHandlers},
{noreply, NewState};
However I can't use this because SASL reports are generated for normal
terminations. Instead I have to handle the normal expected termination
case in the same way as the unexpected case (i.e. inefficiently):
handle_info({'EXIT', Fsm, _Reason},
#state{handlers = Handlers} = State) ->
Fdel = fun(_F, {Key, Pid, _Iter}) when Pid == Fsm ->
Key;
(F, {_Key, _Val, Iter}) ->
F(F, gb_trees:next(Iter));
(_F, none) ->
none
end,
Iter = gb_trees:iterator(Handlers),
case Fdel(Fdel, gb_trees:next(Iter)) of
none ->
{noreply, State};
Key ->
NewHandlers = gb_trees:delete(Key, Handlers),
NewState = State#state{handlers = NewHandlers},
{noreply, NewState}
end.
I would be willing to implement a patch if the team agrees with the
approach.
-Vance
(*)
http://github.com/vances/radierl http://motivity.ca/radiusOn Wed, Apr 06, 2011 at 10:40:56AM +0200, Siri Hansen wrote:
} We have now discussed this matter and decided that instead of adding a new
} terminate/1 function we will add a clause to terminate_child/2 and allow the
} child to be pointed out by Pid instead of Id.
}
} The correction will be included in R14B03.
}
} Thank you for the contribution!
}
} Best regards
} /siri
_______________________________________________
erlang-patches mailing list
[hidden email]
http://erlang.org/mailman/listinfo/erlang-patches
Locked
0001-Implement-new-API-function-to-shutdown-a-supervisor.patch (2K) 
