[erlang-questions] state_timeout reset in gen_statem

[Raimo Niskanen]

On Fri, Apr 27, 2018 at 12:38:39PM -0700, Nato wrote:
> I may be muddled, but as far I can observe, my 
> code seems correct, but am getting unexpected
> behavior from OTP 20's gen_statem behavior.
> 
> I have a state that times out in, say, 3 seconds.
> I have a `bump' routine that casts to that state, 
> and subsequently set it again to 3 seconds. In my
> tests, the initial timer holds fast, and despite 
> my debugging efforts, the original timer isn't
> reset as the documentation says is ought to in
> the gen_statem manual (state_timeout() type 
> section) :
> 
>     `...
> 
>     Setting this timer while it is running will
>     restart  it  with the  new time-out  value.
>     Therefore it is possible to cancel this
>     time-out by setting it to infinity.
> 
>     ...'
> 
> Some relevant code as follows:
> 
>     ...
>     waiting({call, From}, {entry, Key}, #session{}=Ld) ->
>         {ok, Ld1} = handle(start_session, {Key, Ld}),
>         Reply     = {reply, From, ok},
>         Timeout   = session_duration(alive),
>         {next_state, alive, Ld1, [
>           Reply, {state_timeout, Timeout, hard_stop}]};
>     ...
> 
>     alive(cast, bump, #session{}=Ld) ->
>         ok = handle(bump_session, Ld),
>         Timeout = session_duration(alive),
>         {keep_state, Ld, [Timeout]};
>     ...
> 
>     session_duration(alive) ->
>     timer:hms(0, 0, 3);

This function returns 3000, which, when returned in from alive/3 as
{keep_state, Ld, [3000]]} is equivalent to
{keep_state, Ld, [{timeout,3000,3000}]}.

That sets the event timeout in addition to the running state timeout.

With "this timer" the manual understands that there are different
"this timer"s: The event timeour, the state timeout, and each
generic timeout are all separate timers.

You should return
{keep_state, Ld, [{state_timeout,session_duration(alive),hard_stop}]}
instead.

/ Raimo Niskanen


>     ...
> 
> Indeed, the new `bump' cast, never resets the
> original, extending the timeout to a fresh 3
> second timeout.
> 
> On a separate note, any best practices for putting
> together test code that juggles timeouts to
> validate such things, would be appreciate.
> 
> _______________________________________________
> erlang-questions mailing list
> erlang-questions@REDACTED
> http://erlang.org/mailman/listinfo/erlang-questions

-- 

/ Raimo Niskanen, Erlang/OTP, Ericsson AB