W3C home > Mailing lists > Public > public-automotive@w3.org > January 2020

Re: VSS Layers introduction

From: Gunnar Andersson <gandersson@genivi.org>
Date: Tue, 21 Jan 2020 17:26:00 +0000
To: Magnus Feuer <mfeuer1@jaguarlandrover.com>, Rudolf J Streif <rudolf.streif@ibeeto.com>, Ulf Bjorkengren <ulfbjorkengren@geotab.com>
Cc: W3C Public Automotive <public-automotive@w3.org>
Message-ID: <0101016fc923bf16-61500858-b667-4577-8e07-8039677278bf-000000@us-west-2.amazonses.com>
On Tue, 2020-01-21 at 16:35 +0000, Magnus Feuer wrote:
> Ack on the problems of specifying overlay rules as metadata. Let's
> drop that.
> 
> However, we still need to be able to wipe subrees and completely
> replace them with an overlay. 
> 
> We could do it from the command line, which is not an ideal solution
> but would work:
> 
> 
> $ apply_vss_overlay.py \
>     --master=VehicleSpecification.vspec \
>     --delete "Vehicle.Cabin.Infotainment" \

OK, so this specifies a Node path as an input parameter, rather than
listing in a layer file.  It is a new and different way, but perhaps it
solves the specific need you see.

The example also implies something (I think?) which is that addressing
the node of branch type will delete the entire tree below that node,
right?  We have had similar discussions before, in which I have
preferred what feels more explicit to me: 
"Vehicle.Cabin.Infotainment.*"

It also implies, I think, that the parameters are processed in order.
First delete the subtree from master, then it is possibly added again
in your "overlay", right?

Just highlighting that the details matter in the end, i.e. they must be
specified.  We need also to split what is part of official VSS
semantics from what might be specific tool behavior.  (But this
exploration seems fine to lead up to that understanding).


>     --overlay=my_overlay.ovspec

As a concept, yes.  In terms of VSS Layers I envisioned each layer type
to have a more telling and specific name than simply "overlay".  A use
case / needs discussion might help clarify.

Otherwise interesting approach but my mind is in a different space
around this so far:  |I'm less at the point of defining the exact tool
usage / API and I think about this from a semantical perspective, i.e.
a "Specification of the Vehicle Signal Specification" perspective,
before discussing tools.  But of course there is some relationship.

- Gunnar

-- 
Gunnar Andersson <gandersson@genivi.org>
Development Lead
GENIVI Alliance
Received on Tuesday, 21 January 2020 17:26:09 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 21 January 2020 17:26:10 UTC