<PatrickLue> https://github.com/w3c/automotive/issues/266
PatrickL: Ulf brought up starting point, settling on which of the existing approaches and then how to build upon it
Ulf: that is a good summary
PatrickL: I was taking a look at
the features and then how we could design a solution
... I want to avoid the loop we have been in
Ulf: my proposal to avoid the loop and comparison if we make a clear decision up front
PatrickL: I wanted to look at it
in the abstract first whereas you want a concrete example in
front of us
... we also have an email from Gunnar for consideration
... yes my role is to try to bring the solution we have forward
to W3C for possible standardization
... if it were well received we have more modules to bring
forward
... I could explore getting permission internally about
bringing those forward if that helps the group's
consideration
... I do not want the group to be restricted on either solution
but bring forward a best solution
Gunnar: I like stepping outside
the box as well
... I asked to compare the data sets, see how we cover the
signals not the format itself
... my point is not to force one order or solution
... if the group agrees to have a poll I will participate
https://lists.w3.org/Archives/Member/member-automotive/2018May/0021.html
Joakim: the proposal is to make a decision on the three options in front of us, use VISS, ViWi or from scratch
Gunnar: we need to discuss the
different possibilities before voting
... one possible way forward is stay with VISS/VSS for signals
and ViWi for the other domains but that didn't see to go over
well
Ulf: if we were to choose VISS as
the starting point, I had some concrete suggestions for what to
add
... the data model in VSS should be enhanced to have more
object like nodes like ViWi uses. enhancing the data model and
ViWi type of query mechanisms we could then adopt the media
domain from ViWi more easily
... it would be possible to go your route after that
exercise
Laurent: what I liked about the
feature exercise that PatrickL is bringing us through is we
collectively express what we need
... we can free ourselves from both and take features from each
as we discuss. tree model and query for example
... I do not have strong ties with either at this point
Joakim: what is your area of interest?
Laurent: sensors/actuators. I see
the reasoning of a tree structure
... we have two sides since they are influenced in what we
know. I do not see them as very different
... we are saying some of the same things with different
vocabularies
Joakim: I also wasn't part of the
group earlier
... there are car experts here and legacy with depth and
details, trying to cover it would be a long exercise
Paul: a few things expressed
resonated with me
... people have been favorable or critical of individual
features from each
... we have two approaches and set out to solve the problem in
slightly different ways
... this group now has more focus and perspective
... going over what we want, whether feature is from either
approach or independent
... I want to see coalescing of the group
Ted: I like the idea of continuing this feature exercise and then can construct a poll
Paul: if modeling the data with both structured and flat views. @@
Ulf: agree we can make a better decision if we are better informed
<Paul> Gunnar proposed looking at different data and approaches
<Paul> He suggested getting various sets together for comparison
PatrickL: sounds like people are
open to having features based discussion in abstract, avoiding
concrete influences of the two approaches
... people are welcome to present alternatives to the group but
I think we would be better focused on a single task for now
Gunnar: we will be influenced from our collective past experiences
Paul: the feature topics should come together, we would need to avoid looping discussions on them
Laurent: that is what I want, for
us to be influenced from past experiences. as you say we are
not starting from scratch
... this would be a good exercise
Gunnar: especially if we adhere to goal of retaining backwards compatability
Paul: how should we go forward?
Joakim: I like the idea of identifying the leaves first, certain core signals
Paul: we did that some time ago, where we defined core signals with GM and JLR
Gunnar: did that feed into VSS?
Paul: it did
... I would need to find the artifacts in wiki etc
Gunnar: we can have a main track we are trying to do with the group as PatrickL said but allow people to attempt alternates and bring back to the group
Joakim: I see the data model as key
Laurent: tell us why?
Joakim: it is one way to move forward
Laurent: what are the features you want in the data model?
Joakim: having a URL to provide a web developer the vehicle speed
Laurent: that is a valuable
feature and to have in mind when creating such an API
... providing the developer with the features they want. you
provide us a why in a feature list and explore it
Joakim: who is this new charter
for? the earlier one was from car experts
... we are also getting into other modules. we do not want to
create features that developers don't care about
PatrickL: it seems we are in
agreement to continue on this exercise, capturing ideas and
work
... while we can have parallel activities, I want clear
features list
Paul: agree, we just want to
avoid losing thoughts and encourage people to contribute them
to the wiki
... I want everyone to be responsible for capturing ideas and
contributing them to the wiki
PatrickL: I have some ideas for furthering this in an email discussion until next call and in the meantime please feel free to contribute to the wiki
[adjourned]