Send an email to inputs-and-semantics-wg@lists.trustoverip.org to request a calendar invite (you can subscribe to the mailing list at lists.trustoverip.org).
Guiding Goal: The granularity of OCA. What might the next version of OCA look like?
Navigating the revamped wiki page: Inputs and Semantics WG (Paul—10 mins)
Human-readable schema specifications (Steven—10 mins)
News from the Operations Team
From the chat:
Mike Bennett to Everyone (6:27 PM)
You can't rely on human-readable labels as a reliable source of meaning. Need some formal semantics.
Neil Thomson to Everyone (6:29 PM)
Need schema path identification scheme.
This is used in dimensional data models as the "schema base" identification model
Steven Milstein to Everyone (6:30 PM)
If you consider systems that support Internationalization (i18n), attribute names, like labels, messages or any text would have unique IDs(keys)
It’s time consuming from a human-readable point of view, if you’re only concerned with one language.
Can the schema base have “required overlays”?
Robert Mitwicki to Everyone (6:43 PM)
yes we call it core overlays
Burak Serdar to Everyone (6:46 PM)
I suggest you change the schema base examples to use hashes as attribute names, so they don't look like arrays.
Neil Thomson to Everyone (6:46 PM)
Would agree on the base data - which is the storage format. Otherwise "type" is really formatting convenience (e.g. time of day, height, weight)
Carly Huitema to Everyone (6:47 PM)
The risk of hashes is it is really hard to type or compare. I agree attr-1 suggests an array, but it is also easy to compare.
Paul Knowles to Everyone (6:48 PM)
Thanks, Burak. Agreed. We can go with hashes as attribute names.
Salvatore D'Agostino to Everyone (6:49 PM)
depends on the size of the name space.. collisions concerns
identifier space..
Burak Serdar to Everyone (6:50 PM)
Are attributes globally unique? Or are they only unique within the schema it is used in?
Robert Mitwicki to Everyone (6:51 PM)
the idea was to have something global unique
as soon as we can capture the context
Burak Serdar to Everyone (6:52 PM)
I suppose that can work, but then you get into a registry of attributes, who governs that, etc. I suggest the only requirement be that they are unique within the schema, and schema.attr globally identifies it
Salvatore D'Agostino to Me (Direct Message) (6:53 PM)
consider signed object vs. hash
Mike Bennett to Everyone (7:02 PM)
I concur this can be used to present and review schema details in the different OCA layers.
Participants (Name / Location / Time zone / Affiliation):
Leadership positions: