Closed garejon closed 3 years ago
Hi @garejon
Thanks for posting this issue. If you are using the smapiClient.simulateSkillV2(), it should be able to keep the session open. Our VSCode extension also use this api to achieve the skill simulator feature and it support multi-turn dialog. This is the smapi api behind the simulateSkillV2 function: ref. As it mentioned, developer could keep the session open by setting session.mode = 'DEFAULT'
.
Thus, when you use the simulateSkillV2 function, could you please try to set the session property such as here, and see if the issue still exist ?
Thanks, Shen
Hi @ShenChen-Amazon,
I am working on the same project with @garejon I can confirm we are using the DEFAULT
session mode, however our skill relies heavily on the Audio Player, could this be the reason why we can't do multi turn interactions ? We are trying to tests a simple play -> pause -> resume scenario and when we try to pause
or go next
for example with the simulator, the session is gone and the commands are not recognised.
Thanks a lot for your quick response.
@leotorres-16 @garejon ,
Thanks for the confirmation. I reached out to service team asking if this API support audioPlayer use case, the answer is unfortunately this simulate api right now doesn't support this use case. When the simulate api is invoked, it will auto register device for testing the skill. However, the auto registered device so far doesn't have certain capabilities like player/timer/alarms. Thus, you may need to use a real Alexa device for testing the skill for now.
Sorry for the inconvenience. I think it's a really good feature request to the service side. Would be great if you guys could log this feature request at Alexa Forums and vote for it. I will close the issue on SDK side, please feel free to reopen if you have any questions :)
Thanks, Shen
I'm submitting a...
Expected Behavior
Multi-turn for SMAPI SDK: Currently a request to an Alexa skill can be simulated, but it is not possible to simulate two requests that are part of the same skill usage. For example, say I am listening to a some music, and want to say 'next' to go to the next song, SMAPI will act as if 'next' is being said as part of a different session. As if you've closed the skill, and then said 'ask [skill name] for next' or just 'next'.
Current Behavior
Each request sent to the skill via SMAPI acts as if the skill has been closed and is being reopened, instead of a continuous session, where the token sent to the Alexa skill would be preserved between calls.
Possible Solution
Steps to Reproduce (for bugs)
This is in reference to using the following class and its functions: const smapiClient = new Alexa.StandardSmapiClientBuilder() smapiClient.simulateSkillV2() smapiClient.getSkillSimulationV2()
Context
Without this capability, comprehensive end-to-end testing is not possible.
Your Environment
Node.js and NPM Info