Meeting minutes
VISS review
Ted explains how we are conducting review
Ulf: we have an open issue for minor editorial changes and linked a Google doc where we have entered areas we want to address
Ted: please check the document before suggesting others in github #364, it is a catch-all for editorial suggestions, areas that need elaboration
… more substantive topics the group should debate are candidates for separate issues
Peter seeks clarification and Ulf elaborates how we are trying to organize and funnel comments
Ulf: we have some thoughts already on areas to improve
… Ted has some ideas on rewriting introduction/abstract
Ted: yes, history, motivation and use cases - welcome input
Peter: you should perhaps explain what VISS stands for :)
Ulf: status of document is boilerplate
Ted: correct
Peter: for both Core and Transport, I dislike using capitalization how we do
Ulf: example?
Peter: Client, WebSocket, etc
Ted: pet peeve as well is CamelCasing but terms we name are also typically have initial character capitalized
… will look at conventions/rules to be consistent and we can look to minimalize
Ulf: in access control model it makes sense to have clear label for Client
MagnusG: in cases we have special meaning, shouldn't we link to the definitions?
Ted: we are doing some and identified need to do more, within reason especially as we are split between two documents
Ulf: conformance is boilerplate, terminology is thin
… Ted and I have discussed using just VISS instead of VISSv2
Ted: we need to have more terms and maybe more definitions here
Gunnar: I would prefer hosted at for VSS
Ulf: for data model we refer to VSS2 but don't go into details, wonder if we should have more
Ted: was thinking at least an excerpt, high level summary on VSS but alright if others oppposed
Gunnar: question is where to draw the line
Ulf: Service discovery is part of metadata requests, my proposal is to remove this section or move down
… Interface chapter defines all methods, introduction perhaps a bit thin
… we state we support HTTP and WebSocket but allow for others, making clear they need to support read and update
… responses given for each method, please read and like other sections please point out anything unclear
Peter: on subscribe, its use of notification is a bit confusing and should be explained a bit more not in 5.1.3
… 'the notification messages are sent by the filtered data' could be rephrased
Ulf: if you have thoughts on alternate wording
Peter: have and provided
Ulf: great, yes in issue and we will put in our todos document
… for error responses we leave message part as May since not all tranports may support it
… we will add a privacy subchapter under security, elaborate on use of encryption, best practices and summary of/link to access control
Ulf: definitions are out of sync between core and transport, we need to harmonize them
… filtering is a big new section compared to VISS v1
… I am giving thought on the layout or if there are things we can do to make it more readable
… again welcome ideas
… multiple signal requests in filtering chapter required some extra thought, expectations to set for eg error handling
Gunnar: I also encourage you to look at your implementation and the spec for sections that are difficult to understand
Ulf: all of this is implemented
Gunnar: others do not benefit from the internal knowledge you have, my point is to reflect on difficult areas and a fresh perspective
Ulf: we have a separate issue on #metadata-type as #anchors are not sent via HTTP, used in page post processing
… will use query instead of fragments
… we have static and dynamic metadata available
[notes a placeholder ***anymore***, Ulf encourages people to give it thought]
Gunnar: I am not convinced about this chapter. regarding use of fonts, casing, there could be something done to make it more obvious
… I note eg metadata-type is unclear if it is a type that would be entered or that you should use 'metadata-type' - it is a bit ambiguous
Ulf: access control is a rather large chapter, we have discussed this chapter in detail previously so in interest of time for this call stay high level. do give it more detailed read offline
… we want to attract security experts for review as well
Ted: the spec process will prompt a formal review for certain, we can request one before we trigger one as well
Peter: I'm not sure layering is adequately defined
Ulf: it is still being discussed in VSS so perhaps we should omit it
Gunnar: I asked on other call for feedback on basic concept, moving from a readme to actual VSS documentation
… suggest keeping an open issue, there is a chance it will be ready in time
Ulf: we will go into Transport later, please continue reading and providing feedback