Platform Update 1.1.0 (0.61.1) - 25/02/2019

Interesting, thank you very much @ozie! Pray tell, where did you find these files?

Hi all,
I’ve tried the update on my environment (1 Master and 2 Satellites) today and I’m wondering about the result:

  • connected to each device in a individual putty session. sam status: Connected to device
  • Master: result of sam status after “sam update” snips-service… 0.61.1
  • both Satellites: result of after “sam update” :

Finished updating the platform
sam status: snips-service… 0.60.10

Master is running on Debian VM, Satellites ar PI3b+.

What went wrong here?


in the snips install directory there is a folder called snips-actions-templates and in there are templates to use

Can someone from SNIPS confirm that they are aware of all these problems and are fixing them ?

Console intent matching broken and generated python file from snippets broken.

1 Like

Hi @KiboOst,

We are currently investigating issues and trying to figure out the problem.

We do agree we certainly need to improve our changes control and end to end tests strategy as we move forward with more features.

Thanks again for your feedbacks and understandings.

1 Like

At this point it sounds like there are a lot of problems with this update. I held off upgrading because I was worried about this happening. Are the developers recommending people go ahead with an update or is it completely borked (for homeassistant users for example.)

Should the update be taken offline? Is it possible to rollback to pre 1.1 functionality?


It depends on your use cases and tech level.

If you are using your platform in production, meaning “every day use”. The best would be to wait a bit or make sure you have backed up your system before upgrading.

Also breaking releases are more likely to bring issues than the one bringing only minor updates.

If you are using your platform for exploration, regression are less likely to impede your use cases.

Regarding rollback, it’s best to move on and fix the current issues. The platform is still usable for many users.

Regarding Hey Snips V4, On our side our tests shows a lots of improvement on the recall and false positive performances. We hope you’ll find the same results.

@ozie Thank you very much!

Let me know if I can provide more info, test etc.
Actually 50% of my snips installation is down …

There are currently two known issues

both of the above issues will be fixed for the 1.1.2 release.

@Val What about nlu not catching right intent anymore ?

Intents are not recognized correctly anymore

And what about Google ASR which does not work in latest version?

I really would like to help by providing more info, but for now I see that nobody cares :confused:

We are looking into it - we care :slight_smile:

1 Like

Another problem is the missing InstantTimeValue in hermes_python.ontology. Another thread mentions this, but I haven’t seen any response. The snips-skill-owm is broken, e.g., because of this. A bunch of apps available from the store are broken because of this, too (some because they use snips-skill-owm).

Hi @chrisb,

I’m looking into it this morning.

Keep you posted

Can you provide me with a stacktrace ? I’m unable to reproduce this bug.


As answered on your forum post.
So far, so good on our side. snips-asr-google seems ok.

If you can post on your thread the logs of snips-asr-google -v it could help.

Does someone else in the community encountered an issue on this?

Of course. First, here’s the intent that’s generated:
-----cut here-----
{“sessionId”:"",“customData”:null,“siteId”:“pi”,“input”:“what 's the weather tomorrow”,“asrTokens”:[[{“value”:“what”,“confidence”:1.0,“rangeStart”:0,“rangeEnd”:4,“time”:{“start”:0.0,“end”:0.35999998}},{“value”:"'s",“confidence”:1.0,“rangeStart”:5,“rangeEnd”:7,“time”:{“start”:0.35999998,“end”:0.48}},{“value”:“the”,“confidence”:1.0,“rangeStart”:8,“rangeEnd”:11,“time”:{“start”:0.48,“end”:0.78}},{“value”:“weather”,“confidence”:1.0,“rangeStart”:12,“rangeEnd”:19,“time”:{“start”:0.78,“end”:0.98999995}},{“value”:“tomorrow”,“confidence”:1.0,“rangeStart”:20,“rangeEnd”:28,“time”:{“start”:0.98999995,“end”:1.89}}]],“intent”:{“intentName”:“searchWeatherForecast”,“confidenceScore”:0.9897657},“slots”:[{“rawValue”:“tomorrow”,“value”:{“kind”:“InstantTime”,“value”:“2019-03-06 00:00:00 -05:00”,“grain”:“Day”,“precision”:“Exact”},“range”:{“start”:20,“end”:28},“entity”:“snips/datetime”,“slotName”:“forecast_start_datetime”,“confidenceScore”:1.0}]}
----- cut here -----

Now for the stack trace:
----- cut here -----
[owm][err] Traceback (most recent call last):
[owm][err] File “_ctypes/callbacks.c”, line 315, in ‘calling callback function’
[owm][err] File “/var/lib/snips/skills/snips-skill-owm/venv/local/lib/python2.7/site-packages/hermes_python/ffi/”, line 61, in convert_arguments_when_invoking_function
[owm][err] return func(hermes_client, *parsed_args)
[owm][err] File “./”, line 146, in searchWeatherForecast
[owm][err] datetime = getDateTime(intent_message)
[owm][err] File “./”, line 83, in getDateTime
[owm][err] if isinstance(tmp, hermes_python.ontology.InstantTimeValue ):
[owm][err] AttributeError: ‘module’ object has no attribute ‘InstantTimeValue’
----- cut here -----

This is from stdout of snips-skill-server, and in the hope of making the stack trace a little easier to read, I trimmed some from the beginning of each line.

1 Like

I think this was an intentional change, in hermes-python 0.3.0. The other thread suggests that is the case. I’ve submitted a pull request at to fix this. Maybe it’ll be looked at and merged, since there are a couple of published (on the snips console) weather apps that rely on this code. :slight_smile:

Thanks! It looks like the pull request was merged. I’ve submitted another pull request for a similar problem that someone else previously reported – if you invoke that same skill without any slots filled, e.g. you just ask for the weather without specifying time, date, location, etc, it’ll crash.