CODEWITHSUNDEEP
concurrencyerlang
mike10010100
mike10010100

Reputation: 163

Erlang: Child process execute normally until received "die"?

I'm wondering if it's possible to have a child process continue normal operation until it receives a "die" message from a parent process.

Here's the child process's code.

-module(sensor).
-export([start/1]).     % must also export function that will be spawned


%%  create second thread, pass self() as argument so it knows how to reply
%%  exchange messages with second thread until sum=0
start(SlaveNr) ->
    %this part is not right. It waits for die and does nothing else
    receive
        die ->
            io:format("slave ~p: received die~n", [SlaveNr])
    end,
    random:seed(now()),
    Measurement = random:uniform(11),
    if Measurement == 11 ->
        exit(crash);
    true->
        io:fwrite("slave ~p Random # ~p~n",[SlaveNr, Measurement])
    end,
    Sleep_time = random:uniform(10000),
    timer:sleep(Sleep_time),
    start(SlaveNr).

As you can see, it's not executing normally. It's just waiting until it receives a die message. So how can I have it execute its normal function until it receives a die from the parent indicating all child processes should exit?

Upvotes: 0

Views: 110

Answers (1)

Pascal
Pascal

Reputation: 14042

I understand that you want to check the mailbox at each loop, you can achieve this by adding a timout of 0 ms. So the process will check if any received message match to die, if not it will immediately continue.

receive
        die ->
            io:format("slave ~p: received die~n", [SlaveNr])
        after 0 ->
            ok
    end,

It is always possible to stop a process using the exit/2 function: exit(Pid,Reason). In this case you do not have to add any code in your sensor module. If you use the atom normal as Reason, the process will die exactly as if is dying normally by itself.

Upvotes: 1

Related Questions