(Minutes taken by Steven Legg)
Attendees were Hal, Steven and Cyril, but Hal had technical issues and was unable to hear the conversation.
Hal reported that Bill is engaging with the OASIS IT consultant on getting GitHub notifications forwarded to the TC mailing list.
Shortly before the meeting Cyril suggested in the GitHub issues that we should set up a vote on the proposals for renaming XACML. Steven agreed but noted that an OASIS poll would probably only be the voting members and we would want to get the opinions of the non-voting members and observers, which would require an external poll. If there is a clear trend from the membership at large then we voting members can be guided by that.
Steven highlighted the recently started discussion and example in GitHub on rewriting the core spec to be syntax agnostic. Coincidentally the LEXIDMA TC have just published DMLex Version 1.0 that takes a similar approach. DMLex loosely defines an abstract data model in prose that is used to describe DMLex structures. The DMLex core also has prose descriptions of translation strategies for generating XML Schema and JSON Schema from their structure descriptions though it is likely that the XML Schema and JSON Schema files that accompany the DMLex core were manually created. Steven suggests that because we have an existing XML Schema for XACML version 3.0 that we want stay aligned with as much as possible we would not follow a regular scheme for translating from a core data model into XML Schema but rather handle the translation structure by structure.
Cyril is aware of tools that translate XML Schema into JSON Schema so once we have the XML Schema sorted out we can generate the JSON Schema from that. Steven wondered how things like element substitution groups and particles are handled since they don't have counterparts in JSON Schema. In any case this is a one-off task and we can edit the output where needed.
Steven pointed out that he hadn't yet looked into YAML and what is used in the way of schema. Cyril reported that JSON Schema can be used with YAML. Steven said that in that case we could choose to combine the JSON and YAML profiles into one since they use the same schema language, or we could still have a short, separate YAML profile that refers to the JSON profile for its formal definition and otherwise has examples in YAML.
The rewriting of the core spec to be syntax agnostic will be a major task so Steven is taking the approach of addressing the GitHub issues that involve removing and simplifying parts of the core spec first so that we don't rewrite stuff that is destined to be discarded or significantly changed and we don't have to rewrite the new stuff.
The next meeting will be on Thursday, October 10.