|
|
| Author |
Message |
|
| klacke |
 Posted: Thu Feb 21, 2008 2:05 pm |
 |
|
|
User
Joined: 28 Feb 2005
Posts: 138
|
This is something that has bothered me for quite some time.
There is no way to write a function which calls
gen_server:start_link() that then returns a nice controlled error.
If we return ignore from out init() callback, there is no way
to propagate a Reason.
If we return stop or {stop, Reason} we get all sorts of
bad behavior including ugly printouts in the error log.
Furthermore if we are
a) trapping exits - we get an EXIT signal in our inbox from
a Pid we've never heard of
b) Not trapping EXITs, we die
All I want is to silently return
say we have in myserver.erl
start_link() ->
gen_server:start_link(?MODULE, [], []).
....
init([]) ->
case file:open("FOO") of
{ok, Fd} ->
{ok, mkstate(Fd)};
{error, Reason} ->
{stop, Reason}
end.
It means that myserver:start_link() really sucks. It has
different behavior whether the caller traps exit or not.
The solution is to have
init([]) ->
case file:open("FOO") of
{ok, Fd} ->
{ok, mkstate(Fd)};
{error, Reason} ->
{ignore, Reason}
end.
Index: gen_server.erl
===================================================================
--- gen_server.erl (revision 15284)
+++ gen_server.erl (working copy)
@@ -31,6 +31,7 @@
%%% ==> {ok, State}
%%% {ok, State, Timeout}
%%% ignore
+%%% {ignore, Reason}
%%% {stop, Reason}
%%%
%%% handle_call(Msg, {From, Tag}, State)
@@ -285,7 +286,12 @@
exit(Reason);
ignore ->
proc_lib:init_ack(Starter, ignore),
+ unlink(Parent),
exit(normal);
+ {ignore, Reason} ->
+ proc_lib:init_ack(Starter, {ignore, Reason}),
+ unlink(Parent),
+ exit(normal);
{'EXIT', Reason} ->
proc_lib:init_ack(Starter, {error, Reason}),
exit(Reason);
/klacke
_______________________________________________
erlang-patches mailing list
erlang-patches@erlang.org
http://www.erlang.org/mailman/listinfo/erlang-patches
Post recived from mailinglist |
|
|
| Back to top |
   
|
| mdempsky |
| Posted: Thu Feb 21, 2008 5:23 pm |
|
|
|
User
Joined: 24 Sep 2007
Posts: 118
|
On 2/21/08, Claes Wikstrom <klacke@hyber.org> wrote:
> This is something that has bothered me for quite some time.
> There is no way to write a function which calls
> gen_server:start_link() that then returns a nice controlled error.
I can imagine uses for that, but I think crying "gen_server broken !!"
over this is a bit sensational.
_______________________________________________
erlang-patches mailing list
erlang-patches@erlang.org
http://www.erlang.org/mailman/listinfo/erlang-patches
Post recived from mailinglist |
|
|
| Back to top |
|
| klacke |
| Posted: Thu Feb 21, 2008 10:03 pm |
|
|
|
User
Joined: 28 Feb 2005
Posts: 138
|
Matthew Dempsky wrote:
> On 2/21/08, Claes Wikstrom <klacke@hyber.org> wrote:
>> This is something that has bothered me for quite some time.
>> There is no way to write a function which calls
>> gen_server:start_link() that then returns a nice controlled error.
>
> I can imagine uses for that, but I think crying "gen_server broken !!"
> over this is a bit sensational.
I know - I chose the mail title mostly for fun. Nevertheless gen_server
as it is now is not suitable for transient gen_servers where the
init() cb may return an error.
If this is your usecase it is better to roll your own server - which
is boring.
This patch can be made completely backwards compat and should be
added in some form.
/klacke
_______________________________________________
erlang-patches mailing list
erlang-patches@erlang.org
http://www.erlang.org/mailman/listinfo/erlang-patches
Post recived from mailinglist |
|
|
| Back to top |
|
|
|
All times are GMT
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You cannot download files in this forum
|
|
|
Digg It
Del.icio.us
Reddit
Facebook
Stumble Upon
Technorati
