See also: IRC log
<kaz> [ Dave_Raggett, Dan_Romascanu, Daniel_Peintner, Darko_Anicic, Katsuyoshi_Naka, Matthias_Kovatsch, Michael_Koster, Pablo_Ccori, Toru_Kawaguchi, Yingying_Chen ]
OK
<kaz> michael: ok
Sebastian: 2 topics today
... Landscape update and TD review from plugtest
Dave: add late binding concept
<dsr> and integrity constraints
<kaz> TD tech landscape
Sebastian: Shares screen, Landscape is complete but for one entry
<kaz> [ Dave_Raggett, Dan_Romascanu, Daniel_Peintner, Darko_Anicic, Katsuyoshi_Naka, Matthias_Kovatsch, Michael_Koster, Pablo_Ccori, Toru_Kawaguchi, Yingying_Chen, Johannes_Hund ]
Johannes shows the main Tech Landscape HTML page
<kaz> main tech landscape
Johannes: can convert .md to HTML and add
Sebastian: will integrate the TD landscape into the HTML - make a links to the detailed cros-TF
Sebastian: Next will update the CP web page with the current TD status
Sebastian presents Graphic illustration of TD
with 6 aspects of TD
Current practices will be updated with this description
https://github.com/w3c/wot/tree/master/current-practices
<kaz> current practices (github.io)
Metadata, Context, Commmunication, Security, Data Type, Thing Interaction
Dave: add dependencies between
things, version management
... discovery + provisioning
... also depende on purpose of the diagram
Sebastian: This is high level not meant to be detailed
Dave shares screen
Dave shows "W3C view of horizontal metadata" with 3 broad categories
Things, Security, Comms
Dave relinquishes screen
Sebastian: Data type is added
Kaz: What about time and state information?
Sebastian: yes, time and state should be included in the metadata part. can add this also location, FW version, etc..
<kaz> fyi, EMMA 2.0 spec
Kaz: maybe EMMA data format could
be added
... metadata
can talk with the MMI WG guys this week, wait for the update of the document, and add some information to the main landscape HTML next week
Sebastian: TD tutorial
... Discussing new vocabulary for linking security
resources
s/linking secirity resources/ linking resources and security/
Sebastian: redirection server points to a URI that may change frequently
This in in the BCP document
Dave: early binding, late binding
of data models
... Things have properties, events, and actions
... these may be bound early when object is constructed or late
binding at runtime
... also long running actions, would be natural for these to
return a "thing" that enables interaction with the running
action
... this is enabled by late binding
... URI can point to instance, or class, or a late bound
instance of the property
Sebastian: this could use the link "href" to bind
Dave: name could be bound to URI
or an explicit URI or could be a "promise" term to be resolved
later
... an example is a long running action like turning on a
dishwasher
... or result could be one of a set of things but you don't
know the specific
... for a long running action you know to expect a result but
don't know the result
Matthias: concept is very similar to HATEOAS, follow a link and return a "future" which represents a future result like a promise
Dave: well known concept around for decades, need some new types to represent the future or resolution of the promise
Matthias: This is the same
problem as in deferred discovery
... you know a result will be available but need a placeholder
until the value representation is ready
Darko: dynamically typed languages don't need to know the type on resolution
Dave: still may need to know the class of thing that was returned
Matthias: you need to resolve the details of the response in the deferred response but need to know the general class earlier
Dave: allows expressivity in describing a thing by reference
Johannes: could compose descriptions dynamically based on layered capabilities like on/off and light control
Matthias: this is all part of TD, can use for generic templates and for detailed instance information
Kaz: Agrees but late binding is a feature of the TD processor, TD could have classes that define the early information
<dsr> we need to be able to describe no knowledge, partial knowledge and full knowledge for the data model and instance
Darko: this is like duck typing (if it quacks like a duck...)
Dave: Integrity constraints
... looking at IEC standards with integrity constraints for
data models
can use loose constraints like "how long is a piece of string", very flexible
Dave: we may want to consider
this for robustness and resilience, what if data are internally
inconsistent?
... you can detect an eror and throw an exception to the higher
level
Sebastian: Also got a request at the F2F to define this
Dave: in the survey we can look at current practice and past
Sebasitan: Could put it into the BCP document or keep as a workgroup item
Dave: follow programming language
practice
... should at least study and recommend
... need to coordinate with TF-AP
Sebastian: Group review of the TD overview to be added to the BCP document, is this OK with the group?
Sebastian will upload to the web site for review
Sebastian: showing TD tutorial, what is missing is security, communication metadata, etc. should be added to the tutorial
Darko: could we merge Metadata and Context Extensibility aspects?
Sebastian: this is for adding ontologies like SSN
Dave: could talk about "semantic
interoperability" in addition to data type
... people will look for this topic
Sebastian: Other topics?
Dave: centralized or distributed
TD, e.g. DHT, Blockchain
... When metadata are dereferenced, where does it go?
Johannes: see the question as
separate from URI question
... more related to whether we specify centralized or
distributed
Dave: should cover both
possibilities
... not sure where it should go
Johannes: Telehash presentation coming up soon
(Telehash is DHT with blockchain)
Dave: keep aware of emerging ideas
Sebastian: it's a question of efficient discovery also
(TF-DI)
Matthias: DHT for more than just discovery?
Darko: also used for messaging and encrypted decentralized comm
Dave: also relates to security re: oneM2M discussion
Sebastian: AOB?
... will update the BCP and send out for review
... Adjourned
This is scribe.perl Revision: 1.144 of Date: 2015/11/17 08:39:34 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/2 topics/Sebastian: 2 topics/ Succeeded: s/Landscape/... Landscape/ Succeeded: i/add late binding concept/topic: Status tech landscape Succeeded: i/Next will update/topic: Update of the TD content Succeeded: s/DAve:/Dave:/ Succeeded: s/Dave, /Dave: / Succeeded: s/ayes/yes/ Succeeded: s/??/EMMA data format/ Succeeded: s/MMI?/can talk with the MMI WG guys this week, wait for the update of the document, and add some information to the main landscape HTML next week/ FAILED: s/linking secirity resources/ linking resources and security/ Succeeded: s/Sebastian: yes,/Sebastian: yes, time and state should be included in the metadata part./ Succeeded: s/URI</URI/ Succeeded: s/defining/describing/ Succeeded: s/THis/this/ Succeeded: s/getting/got/ Succeeded: s/DAve/Dave/ Succeeded: s/okk/look/ Succeeded: s/re;/re:/ No ScribeNick specified. Guessing ScribeNick: michael Inferring Scribes: michael Present: Dave_Raggett Dan_Romascanu Daniel_Peintner Darko_Anicic Katsuyoshi_Naka Matthias_Kovatsch Michael_Koster Pablo_Ccori Toru_Kawaguchi Yingying_Chen Johannes_Hund Agenda: https://lists.w3.org/Archives/Public/public-wot-ig/2016Feb/0054.html Got date from IRC log name: 17 Feb 2016 Guessing minutes URL: http://www.w3.org/2016/02/17-wot-td-minutes.html People with action items:[End of scribe.perl diagnostic output]