See also: IRC log
<scribe> scribe: tidoust
Johannes: [going through
agenda]
... I note Claes request by mail on documents under
development. Other additions?
... [none heard]
<kaz> Plugfest wiki
Johannes: I think that we have the requirements for the demo, is that correct?
Kaz: Basically, the setting
should be ok.
... The demo lists the demonstrations that will be made on the
first floor.
Johannes: We could do a hackathon to prepare the plugfest. Some exchanges about rooms.
Kaz: The room next to the main
demo hall will be available. That is a room for storage and to
keep devices.
... The meeting planning team suggested to use that room during
the day and in the evening.
... Maybe we can start the preparation at around 3PM.
... The network will not be available on Sunday though.
... Same thing for the AC supply.
... Of course we could bring our devices to the room, but
Sunday will not be a good time.
Daniel: Last time we discussed, there was the question of Internet connectivity.
Kaz: For the plugfest within the meeting room?
Daniel: Yes.
Kaz: We could use wired
connections and routers to create wifi networks if
needed.
... If we can clarify needs, we can talk to the systems
team.
Daniel: And we need to decide who is going to carry around a router.
Kaz: Another question is whether we need Internet connection for the external network.
Johannes: I think we added a column to the table. Soumya does not need connectivity apparently. Let's ask others to fill the column as well.
Kaz: Jonathan filled out that table, need to check with him.
Daniel: Also, if there are too many devices around for discovery, that may be a problem.
Kaz: Right, small local network should solve that.
Johannes: Takeaway is that we
could use that storage room with no time bounding. We need to
decide whether it's Tuesday or Wednesday.
... Which would be better, Tuesday or Wednesday?
Kazuaki: I can do both.
Louay: Also available on both days.
Johannes: Others are not on the call. For Siemens, both days are possible as well.
Kaz: in that case, I suggest to try Tuesday first. If that fails, we have Wednesday as a backup :)
Johannes: Right.
... Any other issue on the topic?
Dave: The request for demos for
monday evening. Is anyone willing to provide a demo for the
developer event?
... See mail I sent earlier on. Just a reminder here.
<inserted> Technology Landscape wiki
Johannes: We have different
consortia that are using different technologies. Several
protocols from the ones we've identified. If people who
contributed to this landscape could also cross-reference the
other IoT efforts so that we know which supports which
model.
... We need some more depth for the technology landscape.
... We had some discussions on CoAP for instance. If we could
complete the Wiki, that would give us more food to make
recommendations.
... Do we have people who can contribute more to this?
<kaz> scribenick: kaz
johannes: coap section
... Dave and Claes?
<tidoust> scribenick: tidoust
Claes: Yes, I can write something
down
... Do you want a separate page for that?
Johannes: No necessarily,
sub-bullets would be fine.
... What would be very good would be to identify criteria that
could be applied to all these protocols, such as availability
for wide area networks.
... That would create a list of questions to answer for all
these protocols.
... We already discussed them for CoAP, but they probably apply
for WebRTC, MQTT, etc.
Claes: OK, I'll see what I can do.
Johannes: Same thing for OneM2M, it would be good to list what protocols they are using.
Michael: I would be happy to add some stuff on IPSO and on OneM2M as well.
Johannes: That would be great,
yes!
... Criteria applied to all of these would be quite good.
Volunteers welcome!
... Regarding the tech landscape, I think it would be good to
detail the different stacks and protocols.
Dave: Do we have an entry for AllJoyn?
Johannes: Not yet. Do we know someone who is involved in that effort?
Dave: No, but Qualcomm is involved in that organization, so we could get in touch with their AC Rep.
<scribe> ACTION: Dave to ask Qualcomm about contributions to Tech landscape about AllJoyn [recorded in http://www.w3.org/2015/10/07-wot-ap-minutes.html#action01]
<trackbot> Created ACTION-11 - Ask qualcomm about contributions to tech landscape about alljoyn [on Dave Raggett - due 2015-10-14].
Michael: They have also something called data driven API, which seems interesting to us.
Johannes: I think that Louay has a proposal here.
Louay: [presenting slides on
WebEx]
... Same presentation as I presented two weeks ago in the
Discovery call.
... Idea of an API for the browser to discover and access
things
... Some of the concepts are taken from my involvement in the
Presentation API.
... Working on a plug-in for Cordova
... The main idea is to define a JavaScript API that allows Web
pages to discover and interact with things. Main point is to
consider security and privacy by design.
... Second point is to abstract from the underlying
protocol.
... The API could be implemented on top of different
protocols.
... Mainly useful for things that are not registered with a
particular server, such as things using Bluetooth
Low-Energy
... [Louay explains the Presentation API, scribe suggests to
look at minutes of Discovery call where this part was minuted
in more details]
-> http://www.w3.org/2015/09/24-wot-di-minutes.html#item01 Minutes of Louay's Thing API presentation in Discovery call
Louay: Focus of Presentation API
is presentation of web content. Audio/video content will be
considered afterwards. We have a Cordova plug-in implementing
the Presentation API.
... Chrome, Firefox implementing the Presentation API right
now.
... The idea with the Thing API is to use the same concept,
because there are common features between the two cases.
... The first analogy is linked to the need to discover
displays/things. The second analogy is linked to the need to
have a communication channel.
... A "ThingRequest" constructor would take a filter object
that describes all things that are of interest for the
app.
... When the app starts the API, the user agent discovers the
things that match the filter. The user selects one thing
(possibly more than one) in the list.
... When the web page gets access to a thing, it gets an ID
that it can use later on to access the thing again later on.
This is useful for usability, as starting otherwise would
always display a dialog to select a thing, whereas for things
you may want to give access for a longer period of time.
... Then the Web page gets a Thing instance, on which it can
read/write properties or call actions.
... The web application will be notified when things are no
longer available.
... We started to implement this as a Cordova plugin, here only
on top of HomeKit, so only available on iOS.
... This is one showcase we want to present at TPAC.
... The goal is also for the Cordova app to serve as WoT
Servient that offers a Websocket server or HTTP server to
enable access to the things from other clients.
... So it's not only an application but also a Service.
... This is the reason why we need to run the app in the
foreground on iOS.
... I received feedback from Francois and Dave.
... Dave's comments are centered on JSON-LD. I don't think they
apply to the API itself.
... Francois wondered whether selection of multiple things
would be useful (in the Presentation API, only one display may
be selected). I think that's a good idea for things. For the
API, instead of one Thing, the application would simply receive
an array of Things.
... Second comment was about how browsers may handle a large
number of things that they discover.
... I think the API is more applicable to
sensors/actuators/things that are not registered on the
network, for home automation. That's the main goal of this
API.
... Last comment was about the dialog display being shown to
the user each time the thing is first accessed. The ID
mechanism that I described would give the app the rights to
access the thing later on without prompting the user
again.
... The Generic Sensors API addresses sensors that are
available on the device where the browser is running. Here it's
more related to discovery new things and interacting with new
things.
... In the Generic Sensors API, there is no discovery. You can
directly check whether a sensor is available.
... Questions? Comments?
Johannes: Can we view the API in two parts, the native selection box and the Thing API itself?
Louay: Yes.
Johannes: Then there is the ability to expose the discovered Thing as WoT Servient. Is that part of the API?
Louay: The API addresses the
first two parts. To offer the thing to other applications, I
will implement another layer. This is really for the plugfest,
not part of the API.
... All the information you need will be in the Thing
description file
Johannes: The reason for my
question is that this would enable Thing-to-Thing communication
in a very unified way.
... This would provide a way to discover things, wrap them for
exposure, and expose them to other things.
Louay: Yes. You will need
specific protocols for discovery and controls, but exposing
could use existing standards more easily. The Thing API I
propose is not a solution for all the features discussed
here.
... We need other parts for registering things. Not part of the
browser API, I think.
Johannes: OK, but you form a
proxy for the thing.
... I wonder why the split between the two parts.
... Risk of fragmentation depending on protocols used.
Louay: Yes. That is a problem with the Presentation API as well. From an application perspective, you would be able to use the same API but may not end up with the same list of available displays depending on protocols supported by the underlying web browser.
Johannes: If you were to implement a thing such as a lamp. You would have another API to expose a thing. Why not have only one API.
Louay: I agree. This is a basic idea. If you can inject your ideas to extend my proposal, I'd be happy to discuss that next time.
Dave: Just a question about the style for accessing things. To some extent, I think this should be language-independent. Also the question of synchronicity as developers will be more expecting synchronous properties for instance.
Louay: If I set a property and then call an action after setting the property. If I don't wait for the property to be set, how can I be sure the action takes place at the right moment?
Dave: Well, you may not be the only one accessing the thing in any case. There may be something behind the scene.
Louay: You propose something such as "thing.colorTemperature = 42"?
Dave: Right, for devs, I think it
makes sense.
... One question is to what level we need to standardize this
API. We need to define a data model, sure. What else? Open
question.
Louay: We could use the event on Monday to get feedback from developers for instance.
Dave: As an IG, we should explore
different options, I think.
... I need we need to get practical implementation experience,
so congrats for this work. The plugfest is a nice experiment.
We need more of that to point out these issues.
Johannes: Would it be possible to upload the script examples in your slides to the GitHub pages?
Louay: Yes, I will do it.
<kaz> scribenick: kaz
tidoust: tx for mentioning all my
comments
... please note Generic Sensor API work handles similar
topics
... could converge common APIs
... upcoming draft will be published soon
... TPAC would be a good opportunity to start discussion
louay: Generic Sensor API focuses
on reading data
... while WoT APIs handle accessing actuators
tidoust: right
<tidoust> scribenick: tidoust
Kaz: Unfortunately, DAP will not
have a meeting at TPAC, but Tobie will be there.
... We can probably have some joint discussion on Tuesday or
Wednesday with him and other people from DAP.
<kaz> ad-hoc meeting proposal (Tuesday)
<kaz> Breakout proposal (Wednesday)
Joerg: Maybe we should have a short list of ad-hoc meetings to the IG agenda. Already a couple planned.
<kaz> +1
Johannes: skipping tooling for today in the interest of time.
<kaz> Claes's message
Claes: I had a look at the
documents referenced by the Wiki and I wondered which of them
are still being pursued.
... For instance, the Architecture Model, is that still
relevant?
Johannes: I would like to open
that question for the group.
... [no comment heard]
... During the F2F meeting in Sunnyvale, we made an attempt to
bring different views for the Web of Things model and
architecture together.
... However, we did not reach consensus to draw this into one
picture.
... It's not very clear to me how we can move forward.
Michael: I would suggest that we
list documents that have clear group consensus and that we
maintain.
... Not doable overnight, but that's a question I got as
well
Dave: Right, by charter, we also
need to publish reports that represent consensus of the
group.
... We need to make it clear what parts are work in
progress
... I think it's a question for the group as a whole.
... Something that the Chair and TF leads could work on for
preparation for the F2F in Sapporo?
Joerg: The most pragmatic is to discuss this by email. Whether the Wiki is fine, where we put parts which have consensus, etc.
Claes: That sounds good.
Dave: The different task forces are working in the right direction, e.g. security
Johannes: Where do we have the reference? How can we ensure that we have a good list of documents?
Dave: Maybe start from the
group's home page with a list of documents on the right.
... And same thing on the IG wiki.
Johannes: Problem is to assess the maturity of documents.
Claes: Wouldn't it be good if task leaders take a stab at it?
<kaz> scribenick: kaz
johannes: I'd get an action item
for that
... to describe the status of the documents
claes: which document for us to contribute
johannes: could list the existing
documents
... that would be feasible
kaz: we can add links to the
existing documents to the "Work in progress" section
... and the moderators can add brief description to each list
item
-> https://www.w3.org/WoT/IG/wiki/APIs_and_Protocols_TF#Work_in_progress work in progress section
johannes: ok
claes: ok
johannes: other TF managers
should also do that
... and make the decision on which to be continued during
TPAC
... have we addressed all your points, Claes?
claes: yes
johannes: tx
... we're getting out of time
... any other business?
(nothing :)
[ adjourned ]