Registering A Callback

Cloud
1

The docs say callback functionality will be ready sometime in March.
Is that still the plan otherwise should we just rely on making TCP connections on the device?

2

Yep still the plan, itā€™s coming very soon. Already developed, in testing now. But in the meantime, try out Server-Sent Events!

2 Likes
3

Thanks for the fast response. Iā€™m using Parse for my backend and looks like they donā€™t support Server-Sent events further more Iā€™m working on a similar project to your pizza ordering button so the callback might be more practical, unless you have a better suggestion?

4

@zach i got really curious and read the blog a few times wonder about thisā€¦

It seems to me that the name for Server-sent-events can be almost anything a user wants.

What if and what if both user happens to be using the same name? eg. ā€˜me/temperatureā€™

Sounds like itā€™s gonna be a rare occasion but i got really curious about it :smiley:

Also, letā€™s say my event TTL is 60sā€¦ And it expired before the event was triggered againā€¦ And someone published with a same ā€˜nameā€™. Wonā€™t i be getting data from that publish? :smile:

Sorry for the silly question cos i guess probably people would sayā€¦ ā€˜use a unique and special name for your eventā€™

Just wondering :stuck_out_tongue:

5

Good question!

It all depends on which endpoint your program is listening to:

e.g:

etc

So even if someone else is using the same event name, you can just filter to your cores, or a particular core. Also, any state persisted from a message you publish is stored only along with your core information, and is only accessible to you.

Thanks!
David

6

That sounds interesting!

Soā€¦ Our events are probably all going under api.spark.io/v1/devices/events whether public or private?

Or anyoneā€™s public event goes to ā€˜All Public Eventsā€™?

7

Just the public events appear in the ā€œAll Public Eventsā€ firehose, if your event is flagged private it will only show up for you in those feeds.

8

got it!

Isnā€™t that going to have some event name clashes in the public when the :spark: core community gets incredibly large which is going to happen? :smiley:

orā€¦ Letā€™s say 100 people uses ā€˜temperatureā€™ public eventā€¦ i can filter mine with api.spark.io/v1/devices/events/Temperature ?

1 Like
9

Hi @Dave

First off, this is great and works wonderfully!

I do have a question thoughā€“when I run a curl command to listen to the events, I get the event data as expected, but I also get occasional carriage returns between event data, sorta like a keep-alive is running. This is expected, right?

And to who ever is publishing ā€œTemperature/Deckā€, wow 3 degrees F is cold even by Boston standards!

10

@kennethlimcp ā€“ totally, api.spark.io/v1/devices/events/Temperature would filter to just events from your cores starting with "Temperature.

@bko - Thanks! That would be my deck, it is cold! Also, yes, the Carriage Returns are an intentional keepalive, they should come in every 9 seconds when there is no other traffic.

11

Ah iā€™m still really curious and having some hands-on makes me pick things up faster :smiley:

So @bko could you share how you do the curl command and how to you know about ā€˜Temperature/Deckā€™ events? :smiley:

I have a python script which uses the ā€˜requestsā€™ module when i tried to help test the CFOD issue to perform HTTP requests.

Also, i modified it slightly to check my access_token and delete etc.

Thatā€™s like the most i ever did with python and API so i definitely need you veterans to start me off with new feature!:wink:

Iā€™ll put up a Tutorial in return. haha!

12

Hi @kennethlimcp,

One thing thatā€™s nice about these endpoints are that they use GET requests, so you can just open them in a browser window, if youā€™re in chrome you have to ā€œview sourceā€ on the page to see whatā€™s coming across:

view-source:https://api.spark.io/v1/events/?access_token=your_access_token

13

@Dave THATā€™s REALLY FUN :smiley:

Seriously i always loved microcontroller and the :spark: made it more fun!

1 Like
14

What happens if a user accidentally calls the Spark.publish() too frequently?

15

Hi @kennethlimcp

I just did what David suggested and used:

curl -H "Authorization: Bearer <<hex number here>>" https://api.spark.io/v1/events/Temperature

You to replace the entire <<hex number here>> part with your Auth Token.

Interestingly this finds events who names begin with Temperature like Temperature/Deck and Temperature/Basement (go in the basement David, itā€™s much warmer! :smile:) but it does not find spark-hq/Temperature. I think that is a good way for this to work with hierarchical levels.

1 Like
16

Then there will be a lot of events! :slight_smile:

17

:smiley: Is there something wrong here which causes a compile issue?

String data = 'ms';

void loop() {
    
        Spark.publish("Core_Uptime/Singapore", data);
18

Yup! in C++ / arduino coding land, 'a' single quotes denote a single character, and "abc" double quotes denote a string. So you want to change 'ms' to "ms"

19

I would recommend starting out trying to not publish more than say, 5-10 events a second, if you send too many too quickly you may knock your core offline :slight_smile: (I suspect itā€™s possible to flood yourself offline currently)

20

Hmmā€¦ Seems like a problem still.

Is this related to this issue?