Reputation: 33
A took a look at http://erlang.org/doc/apps/inets/http_client.html and found the following:
An ordinary asynchronous request. The result will be sent to the calling process on the form {http, {ReqestId, Result}}In this case the calling process is the shell, so we receive the result.5 > {ok, RequestId} = http:request(get, {"http://www.erlang.org", []}, [], [{sync, false}]).
6 > receive {http, {RequestId, Result}} -> ok after 500 -> error end. ok
http:request passes a message to me after calling it that way, but how does it know my process id? As far as I understand, http:request has to do "Pid ! {http, {RequestId, Result}" to send the result to me, but where does it know the value of Pid?
Upvotes: 3
Views: 6028
Reputation: 57658
If you look at the definition of the #request{} record (in httpc_internal.hrl), you will see that there is a field called from. It contains the caller's pid; that's how the server will be able to send a message to the caller later.
Looking at the source code of http module you will see that your call will eventually reach the handle_request function, where the from field is set to self().
Upvotes: 6
Reputation: 11277
You don't -- the calling process has to supply its Pid if it wants a reply. Do something like
Server ! {self(), arguments}
and then the first element of the tuple is the return address.
Upvotes: 2