►
From YouTube: T2TRG Interim Meeting, 2021-06-17
Description
T2TRG Interim Meeting, 2021-06-17
A
Okay
and
now
we
should
be
live,
recording
so
once
again
welcome
everyone
painting
rg
wishing
meeting
here
is
our
draft
agenda.
For
today.
I
will
very
briefly
touch
upon
the
some
of
the
wish
activities
we
have
on
going
and
unplanned
now
in
the
future,
but
the
bulk
of
this
meeting
would
be
either
with
the
focus
on
the
azure
dddl
and
an
sdf
in
their
work
follow-up.
So
we
did
this
start.
The
video
discussion
quite
some
time
time
ago
asking
some
in
progress
on
that.
So
with
I'm
thinking.
A
A
And
finally,
let's
see
if
we
have
time
and
if
we
have
a
actually
sebastian
managed
to
join,
he
said
he
won't
be
available,
at
least
for
the
first
hour.
We
could
briefly
touch
the
topic
of
w3c
web
of
things,
team
models,
which
are,
of
course,
also
very
much
related
to
the
to
the
other
work
that
we'll
be
discussing
today.
A
Return
to
our
roughly
use
locations
around
september,
but
then
there's
also
a
a
ietf
online
meeting
happening
in
july
and
before
the
iedf
week,
that's
usually
a
hackathon
week
and-
and
some
of
us
have
been
taking
part
in
the
hackathon
activities
and
there
was
a
one
idea.
You
know
carson
proposed
that
we
could
also
have
a
wishy
hackathon
activity
there.
So
carson
would
like
to
say
a
few
words
about
that.
D
Yes,
so
we
have
traditionally
been
able
to
make
good
use
of
the
week
before
the
ietf
to
make
forward
progress
on
sdf
on
converters
and
so
on.
So
we
want
to
continue
that
tradition
this
week
and
yeah.
We
still
have
to
to
find
out
who
exactly
will
be
there
with
which
converter
so
the
the
slide
lists
jana's
sdf
to
yank
converter,
which,
by
the
way
now
has
a
website
that
you
can
play
with.
A
Here:
okay,
good,
unfortunately,
it
looks
unlikely.
I
will
be
able
to
make
the
hackathon.
I
will
plan
it
to
be
on
summer
vacation,
but
let's
see
maybe
it's
a
rainy
day
in
finland
and
I'll
manage
to
call
in,
but
I
think
it's
a
very
good
activities
that
we
have
had
in
the
past
as
karsten
said.
So,
if
you
have
a
chance,
I
highly
recommend
joining.
A
B
I
think
there's
probably
a
couple
things
I
can
share
one
is:
we've
now
got
full
support
for
dtdlv2
with
our
current
version
in
in
our
azure
digital
twin
service,
and
so
that's
been
available
for
customers,
and
then
we
we're
getting
support.
That's
coming
online
in
another
one
of
our
services.
Iot
central
one,
interesting
thing
on
the
open
source
side
is
that
we've
actually
got
now
a
set
of
open
source
dtdl
based
ontologies
in
github
for
some
different
domains.
B
We've
got
one
for
smart
buildings
based
on
the
real
estate
core
ontology,
that's
their
our
ontology.
We've
got
that
in
dtdl
available
and
then
we've
got
one
for
smart
cities
and
that's
based
on
now.
I
can't
remember
off
top
my
head,
but
that's
based
on
a
standards
organization
we're
working
with
in
the
smart
cities
domain,
and
then
we've
got
one
on
energy
grids
as
well,
and
so
those
are
all
on
github
and
dtl-based
ontologies
that
are
are
open,
source
and
available
for
use.
D
B
The
licenses
on
the
github,
I
would
have
to
go,
look
at
see
exactly
which
license
it
is
okay,
but
but
in
terms
of
yeah,
so
there's
there's
a
license.
That's
out
there
and
I
can
publish
these.
I
can
put
the
links
into
the
chat
here
in
a
little
bit
great,
but
but
then
they
also
do
accept
some
submissions
as
well.
So
you
know
for
people
working
in
those
domains.
If
you
know
there's
an
area
not
covered
or
something
that
needs
to
be
fixed.
Those
projects
are
also
accepting
submissions.
E
Hi
this
is
bruce
nordman,
I'm
interested
in
what
you
said
about
energy
grids,
I'm
interested
in
well.
I
think
we
can
depose
our
energy
systems
into
wide
area
networks
and
then
local
area
power,
distribution
within
customer
sites
and
then
the
interface
between
them
is
what
you're
doing
cover
all
of
that
or
portions
of
that.
B
So
so
I'm
not
personally
an
expert
in
in
the
energy
grid
space.
I
some
other
people
on
our
team
actually
worked
on
the
details
there
from
what
I've
and
I
don't
even
know,
really
sort
of
the
separation
of
the
different
areas
in
that
domain.
But
it
looks
like
it's
base
it.
It
looked
like
it
was
based
on
from
what
I
remember
like
power
station
substations
lines,
those
kinds
of
concepts
in
in
the
energy
grid
space,
but
I'll
put
the
links
into
the
chat.
E
B
E
B
Yeah,
no
totally
so
we
we're
very
you
know
we're
looking
to
do
more
of
these
things.
I
see
that
michael's
already
actually
posted
the
links
into
the
chat
for
the
different
ontologies
there,
which
is
great
thanks
and
no
we
we
want
to
do
more
of
these
and
and
improve
these.
So
certainly,
if
you
or
you
know
anyone
has
more
more
to
add
or
you
know,
other
domains
that
you
want
to
see
happening,
we've
got
a
a
group
of
people
that
are
actually
working
with
partners
and
other
standards
on
doing
these
ontologies.
B
A
B
I
believe
that
there
is
one
that's
done
by
the
willow
company
that
has
taken
the
the
smart
buildings
one.
I
think
they've
got
this
in
a
public
repository
in
github
as
well,
where
they've
taken
the
smart
building
or
the
other
smart
buildings,
the
buildings
ontology
that
was
based
on
real
estate,
core
and
they've
done.
You
know
added
their
own
customizations
on
top
of
that,
and
so
at
least
from
an
ecology
perspective,
you
can
see
how
they've
used
it.
I
don't
think
that
they've
done
that
they've
published
anything
around
their
actual
solution.
B
Yeah
that
what
they
you,
what
those
ontologies
are
built
for
is
primarily
for
the
use
case
of
use
being
used
with
azure
digital
twins.
So
what
you'll
see
in
those
ontologies?
Is
they
heavily?
You
make
use
of
of
properties
for
all
the
state
information
about
the
different
assets
and
things
equipment
and
that
kind
of
stuff
and
then
and
and
relationships
for
building
out?
A
A
So
first
some
of
our
our
latest
experience.
So
I
I've
been
working
with
a
couple
of
colleagues
in
part
of
a
patriarchy.
My
colleague
has
been
active
here
on
the
implementations
for
conversions
between
stf
and
dddl,
and
as
of
today,
we
do
have
two
implementations,
one
one
for
each
direction.
I
think
for
the
sdf
to
ddtdl.
We
seem
to
be
able
to
quite
complete
conversion
of
sdf
features
that
we
can
express
in
indeed,
l.
Of
course
the
question
is:
are
we
doing
it
right?
A
A
How
much
information
are
we
losing
when
we
go
back
and
forth,
and
how
can
we
import
different
dvdl
models
into
sdf
and
and
that's
why
it
would
be
interesting
to
see
kind
of
a
set
of
dddl
models
that
have
been
done
by
by
someone
specifically
for
dtdl
that
we
could
see
like
how?
Well
can
we
perhaps
express
those
in
sdf,
but
that's
still
an
early
implementation?
A
We
have
been
playing
around
mostly
with
with
the
models
we
already
converted
from
sdf
to
ddl
and
then
back
to
sdf
and
there's
also
a
link
in
the
slides
to
the
repository
some
of
the
examples.
But
we
do
also
have,
as
of
20
minutes
ago,
a
demo
web
service
running
on
a
public
internet
host
that
you
can
essentially
post
your
sdf
model
and
it
turns
it
into
a
dddl
model
and
there's
an
example:
curl
command
that
you
can
use
but
bear
in
mind,
that's
a
experimental
code
and
a
web
service.
F
That
makes
sense
that
using
dtl2
sdf
being
that
dtl
is
an
rdf
that
we
really
focused
sdf
on
how
to
express
device
affordances.
But
sdf
also
has
a
very
generic
object.
Property
model,
so
I'd
be
really
interested
in
following
up
on
what
kind
of
extensions
might
be
needed
to
sdf
to
to
increase
the
information
that
we
can
express
in
sdf
from
something
like.
B
But
before
you
before,
you
go
on,
actually
I
just
wanted
to
comment
on
this
that
I
had
actually
taken
a
look
at
the
conversion
at
least
the
one
way
the
sdf
to
dtdl.
I
didn't
I
didn't
know
there
was
the
the
other
conversion
the
other
way
and
from
what
I
had
looked
at
it
looked
like
those
were,
you
know
fairly
complete
conversions.
I
mean
you
know
fairly
correct
conversions
from
at
least
my
perspective
from
what
I
understood.
A
Excellent
that
that
sounds
very
good.
I
guess
that
yeah.
When
we
look
at
the
two
model,
I
mean
we
see
we're
actually
well
aligned
already
on
kind
of
meta
level.
How
do
we
express
things
so
it
was
relatively
straightforward
thing
to
do.
It
was
more
more
on
the
structure,
whereas
sdf
uses
a
lot
of
nesting
in
the
chase
and
then
in
the
dtdl.
You
use
references
so
that
was
kind
of
the
main
part
of
the
implementation.
A
D
A
A
A
Then
I
think
we
have
a
very
good
easy,
easy,
simple
conversion
between
those,
but
then
what
we
did
realize
that
some
of
the
units
are
not
si
units.
I
guess
the
most
interesting
example
was
this
slug
unit,
and
while
there
is
a
programmatic
way
to
convert
from
from
from
slack
to
to
kilograms,
it's
not
an
si
unit,
probably
it
would
not
be
suitable
for
the
cinema
registry.
A
So
then,
this
raise
a
question
like
how
do
we
express
in
sdf
units
that
are
not
suitable
for
the
cinema
registry
but
are
still
used
by
some
of
the
ecosystems
that
we?
What
we
are
interested
in,
like
like
dddl
as
this
sdf
really
tries
to
be
descriptive
and
trying
to
describe
the
capabilities
of
each
ecosystem?
A
F
F
A
C
A
I
guess
the
radicals
also
would
make
sense
if
there's
a
generic
thing
where
these
are
defined,
then
maybe
using
that
namespace
would
make
sense
if
it
turns
out.
There
is
a
a
unit
that
we
well.
I
know,
maybe
that's
a
question
to
you
at
for
them
on
the
ddl
side.
Do
you,
I
guess
you
get
these
units
from
from
somewhere
and
there's
probably
a
reference
to
external
source
for
these
units,
and
do
you
think
they
would
like
some
qdt,
the
yukum
or
or
something
that
we
could
use
as
the
name
space.
B
B
You
know
not
everyone,
not
not
every
domain
and
not
every
application
is
using
si
units,
and
so
we've
had
requests
for
non-si
units
us
units,
other
types
of
things
like
that
and
so
we've
been
you
know,
been
trying
to
be
accommodating
there.
Recognizing
that
you
know
different
different
applications.
Different
domains
have
these
different
units,
and
so
that's
that's
why
we
have
those
the
the
in
terms
of
the
the
referencing
detail
units.
B
Actually
that's
a
very
interesting
idea
and
we
actually
do
have
a
full
uri
for
each
of
our
units,
and
I
was
just
trying
to
go
find
what
that
is,
but
I'll
go
find
exactly
what
that
that
full
uri
is,
and
so
I
could
imagine
you
doing
some
kind
of
naming
space.
You
know
uri
based
on
that
and
then
it
would
match
exactly
that.
You
know
the
full
uri
for
each
of
those
units
in
dtdl,
which
would
which
would
make
it
be.
B
Yeah,
so
so
all
of
our
uris
are
actually
we
call
them.
Dtmis
digital
twin
model,
modeling
identifiers,
so
they
are,
they
are
uris,
but
they
aren't
urls,
and
so
you
can't
actually
dereference
them
to
you
know
any
you
know
any
actual
resource
and
but
they,
but
they
still
are,
you
know,
used
as
identifiers
for
those
elements
and
then
what
those
do
point
to
as
they
point
to
definitions
in
the
dtdl
meta
model,
which
right
now
is
not
publicly
available,
but
that
is
something
that
we're
looking
at
making
available.
B
B
Yeah,
I
don't
know
the
time
frame
for
that.
It
is
something
that
we're
looking
at.
You
know
we're
looking
at
a
couple
things
with
making
dtl
more
open.
One
is
just
first
to
the
the
you
know:
we
have
a
github
for
dtdl
and
but
we
haven't
really
been
taking
submissions
there
yet
and
so
we're
working
on
that.
D
B
I
think
that
makes
sense
and
what
we
haven't
done
in
our
meta
model,
which
obviously
you
can't
see
yet
because
it's
not
public,
but
even
you
know
our
internally.
We
haven't
done
those
references
back
to
any
source.
You
know,
and
so
I
actually
think
that
makes
sense
for
us
to
consider
on
our
side
for
the
for
the
units
that
we
have
where
there
is
a
you
know
where
there
are
other
definitions
that
we
know
of
is
is
establishing
those
links.
F
B
B
You
know
from
right
now
in
dtdl
you
can
only
specify
them
in
the
model,
so
they
are
fixed.
Some
customers
want
to
be
able
to
adjust
those
at
runtime
based
on
the
the
actual
devices
or
equipment.
That's
installed
because
it
as
they
do
discovery
through
through
those
deployed
devices.
Some
some
devices
are
configured
differently
than
what
the,
but
then,
what
might
be
said
at
the
model
level,
so
we've
got
that
and
then
we've
got
the.
B
A
Okay
seems
to
be
that
we
are
actually
hitting
very
similar
challenges
in
the
unit
area.
That's
indeed,
and
it
was
also
in
in
the
ipso
like
within
the
space.
We
have
been
discussing
this
quite
a
bit
in
the
in
the
in
the
past,
and
we've
done
a
set
of
alignment
that
like
in
them
in
the
ipsworld.
We
also
aligned
with
the
cinema
units,
and
so
that's
I
mean
that
that
was
one
one
way
where
we
actually
get
quite
quite
far
already,
but
actually
the
and
also
the
point
of
having
this
unique
conversions.
A
I
I
fully
agree,
and
the
good
thing
is
that
a
colleague
of
mine
did
that,
for
the
at
least
a
quite
big
part
of
the
cinema
units
and
and
ddl
units.
So
here
you
see
a
fragment
of
the
configuration
file
that
he
did.
That
is
part
part
of
the
program
that
does
the
conversions,
and
here
you
actually
see
that
each
of
cinema
units
has
two
this
unit
and
type
because
in
detail
you
have
this,
what
you
call
semantic
types
that
gives
a
bit
more
bit
more
semantic
information
for
for
those
units.
F
Just
remember
that
that
what
is
it
was
it
sc-41
that
they're
actually
working
on
units
also
in
iso
and
there's
there's
a
proposal
that
has
seven
fundamental
units
from
which
a
bunch
of
other
types
are
derived,
and
I
don't
know
if
that's
really
the
way
we
want
to
go.
But
there's
there's
some
idea
of
instead
of
having
just
a
huge
vocabulary
of
constructed
units,
go
go
to
the
thing
that
can
be
constructed
using
the
basic
units.
D
Yes,
but
it
has
the
disadvantage
that
some
things
that
actually
do
have
a
physical
meaning
reduced
to
one,
and
sometimes
you
just
need
more
information
to
to
see
what
what
this
one
actually
means.
So
do
you
have
liters
per
liters
or
kilograms
per
kilograms,
and
so
on.
So
that's
why
that
seven
dimension
in
space
is
not
always
sufficient,
but
it's
a
good
basis
right.
D
B
Yeah,
I
think
that
would
be
great
I've
been
recently
looking.
I
haven't
looked
at
senml,
so
I
took
a
took
a
note
to
go:
do
that
I've
been
looking
at
qudt
and
ontology
of
measures
units
there
and
trying
to
kind
of
reason
through
and
think
through
how
those
could
connect
to
dtdl,
either
just
directly
enabling
reuse
which,
which
we've
heard
from
some
customers,
that
that
could
be
useful.
B
A
Good
but
then
maybe
we
in
the
interest
of
time,
we
can
now
move
move
forward
and
discuss
briefly,
then,
some
of
the
other
piece
of
missing
information
and
the
way
forward.
So
one
thing
that
we
noticed
when
we
do
it
from
the
vddl
to
sdf
conversion
is
that
in
in
detail.
As
we
discussed,
you
have
these
relationships.
A
We
don't
currently
have
that
same
concept
in
sdf
that
we
already
have
discussed
in
the
standardization
at
the
ietf
that
we
could
be
adding,
adding
that
and
here's
a
draft
example
that
we
discussed.
That
should
be
able
to
express
the
same
information
that
the
ddl
is
expressing.
So
essentially
the
what
you
see
here
contained
in
that
would
be
the
name
element
in
the
relations
target
would
be
target
and
properties
would
be
properties.
A
F
That
we
have
type
link
vocabulary
that
looks
a
little
different
from
this.
We
have
rel
and
href,
and
I
think
I've
brought
this
up
before
and
you
had
some
good
reason.
We
weren't
using
rel
and
href,
but
I
don't
remember
and
then
also
sdf,
are
our
style
is
more
to
use,
singular
and
not
to
pluralize
things,
because
it
sort
of
creates
an
ambiguity
that
a
lot
of
systems
use,
singular
versus
plural,
to
differentiate
things,
and
we
decided
not
to
do
that
and
decided
that
our
preferred
style
was
singular.
B
Good
one
thing
I'm
curious
about
here
is
you've
been
looking
at
the
pdl
relationships
and
what
you
might
want
to
represent
in
sdf
is,
if
there's,
if
you've,
seen
any
shortcomings
there
or
you
know
from
what
you
would
expect
to
be
able
to
do.
If
there
are
some
things
anything
we
should
be
considering
on
the
dtdl
side
as
well.
A
F
G
A
But
but
definitely
I
mean
yeah
so
far
we
haven't
done
any
beyond
like
planning
on
this,
but
I
guess
when
we'll
start
actually
using
this,
then
we'll
know
learn
more
also
whether
the
dddl
current
capabilities
fit
well,
and
I'm
glad
to
hear
that
you
know
you
are
also
interested
in
developing
ddbl
to
make
that
make
it
easier.
So
I
think
that
sounds
also
very
good
follow-up
for
us.
B
Yeah
no,
I
just
asked
because
we
had
made
some
decisions
like
our
relationships
are
one
way.
For
example,
you
know
as
they're
expressed
in
the
model-
and
you
know
there's
you
know,
we've
heard
questions
about
that
and
and
has-
and
it's
still
stuck
with
the
the
one-way
at
the
modeling
level,
doesn't
prevent
you
from
doing
things
like
when
you
actually
instantiate
these
things.
Inside
of
you
know
a
graph
database
or
something
from
being
able
to
query
both
directions,
but
the
modeling
side,
for
example,
was
was
one
way.
B
We've
also
heard
some
feedback
about
the
target
right
now
in
dtdl.
The
target
can
either
be
any.
If
you
don't
provide
a
target
at
all,
it
means
relationship
points
to
anything
or
you
can
provide
a
single
target,
but
we
don't
have
a
way
to
provide
say
a
set
of
targets.
We've
heard
some
of
that
discussion
as
well.
A
D
That's
actually
something
there
is
another
activity
in
the
itf
going
on
to
actually
write
up
a
standard
for
json
path,
which
might
be
a
good
way
to
to
have
a
target
that
actually
doesn't
contain
the
whole
world,
but
is
actually
more
than
one
instance,
and
so
that's
that
may
be
something
that
we
can
feed
into.
The
jsonpath
working
group
of
the.
D
A
B
Yeah
exactly
when
we
have
a
relationship
in
dtdl
say
it's
called
contained
in
like
your
example.
Here
it
on
the
instant
side
like
in
a
in
a
graph
database,
for
example,
you
could
have
many
instances
of
that
named
relationship,
so
many
edges
in
your
graph,
but
if
there's
a
target
specified
depending
on
you
know
the
implementation
or
whatever
you
know,
that
indicates
that
what's
expected
there
at
the
end
of
that
edge,
the
other
side
of
that
edge
is
something
of
that
type.
B
A
Okay,
good
see,
I
guess
here's
a
good
way
forward.
I
mean
we'll
be
experimenting
more
on
what
makes
sense
on
on
the
stf
side
and
then,
when
we
get
more
experience,
we
could
get
back
and
also
see
like
if
there
are
some
things
that
we
would
like
to
express
more.
On
the
dddl
side,
we
could
see
like
how
how
the
best
go
forward
with
those.
A
Good
and
then
for
do
the
other
direction
going
from
a
sdf
to
to
ddl
a
couple
of
things.
We
we
noticed
that
are
we
losing
translation?
One
particle
was
to
swing
sdf.
We
have
this
info
block.
We
have
information,
such
as
copyright
license,
title
of
the
model,
etc.
So
some
of
those
I
mean
like
we
could
put,
for
example,
in
the
description
element
and
your
title
can
go
to
name
element
but
then,
like
copyrighted
license,
they
don't
have
specific
ways
to
express
in
dddl.
B
Yeah
from
the
dtl
side,
it's
something
we've
heard
about
a
little
bit.
One
of
the
other
things
that
we've
actually
been
hearing
is,
that
is
that
it
may
be
more
useful
to
to
provide
some
kind
of
structure
around
a
collection
of
models.
You
know
into
an
ontology,
and
so
we've
been-
and
this
is
you
know
just
this-
is
early
thinking
on
this.
B
We
haven't
really
put
this
on
a
road
map
yet,
but
we've
been
looking
at
the
idea
of
are
there
you
know,
should
we
have
some
way
of
expressing
ontologies
and
then
some
of
this
other
information,
but
I
think
it's
fair
like
maybe
we
need
to
look
at
both.
Can
you
provide
this?
The
copyright
license
et
cetera
in
the
you
know,
at
each
model
level,
and
then,
if
we
were
ever
to
do
some
kind
of
ontology
or
collection
of
models,
maybe
it's
there
as
well.
A
D
D
Yeah
there's
some
something
is
going
wrong
with
my
vdsl
due
to
the
outside
temperature,
so
the
the
manufacturers
uses.
The
description
gives
you
information
about
the
communication
requirements
of
a
device
and
when
you
actually
compose
systems,
then
you
compose
devices
and
you
have
to
compose
the
manufacturer,
uses
the
substance
and
of
course
this
immediately
raises
copyright
issues.
D
So
we
just
have
to
get
this
this
little
detail
out
of
the
way,
because
you
are
composing
devices
from
different
manufacturers,
so
you
have
to
make
sure
that
you
are
even
allowed
to
to
put
these
things
together.
Of
course,
you
can
also
completely
ignore
that
that
aspect
which
we
probably
all
have
been
doing
in
the
past,
but
but
given
that
these
bodies
of
work
are
getting
bigger
and
bigger,
we
probably
should
be
starting
to
think
about
licensing
issues
here,
as.
A
A
A
Is
okay?
Maybe
if
we
continue
then
the
other
one
in
in
in
sdf
files,
for
example,
for
the
properties
and
other
data
elements,
we
can
define
something
more
details
about
what
the
data
shapes
like.
You
know,
whether
it's
value
has
minimum
maximum
or
if
a
string
has
a
minimum
or
maximum
length
or
follows
certain
pattern
or
it's
our
default
and
constant
values.
B
Yeah
again
yeah
from
the
dtdl
side-
yeah
it
that's
not
there.
So
yeah,
you
wouldn't
find
that
our
our
idea
for
that
which
we
haven't
yet
I
guess
made
fully
available,
is
to
is
to
enable
that,
through
our
semantic
types,
so
you'd
be
able
to
enable
you
to
be
able
to
attach
say
on
a
property
in
dtdl,
a
semantic
type
of
say.
You
know
operating
constraints
or
something
like
that.
F
F
D
F
So
there
would
be
a
need
to
do
an
override,
but
actually
you,
if
you
had
an
api,
you
would
really
only
see
the
most
recent
override
value.
You
wouldn't
see
the
the
other
one
or
would
you,
but
you
want
to
include
both
of
those
like
a
fundamental
minimum
and
an
instance
minimum
or
something
like
that
like
is
there.
D
F
F
But
still
seems
like
a
given
instance
of
a
temperature
sensor
in
normal
used.
You
don't
really
need
to
worry
about
the
fact
that
it
can't
ever
go
below
zero
as
a
temperature,
but
mostly
the
usefulness
is
in
knowing
what
the
limit
of
the
scale
for
the
particular
sensor
is.
So
I
don't
know
how.
I
guess
it'd
be
interesting
to
see
how
we
would
you
know
put
that
in.
I
guess
it
would
be
a
different
property
or
it
would
be
a
different.
A
A
F
B
B
B
B
I
guess
we've
been
looking
at
it
for
properties,
mostly
not
for
things
like
telemetry,
but
for
properties
for
for
the
state
information
being
able
to
provide
a
default
value.
We've
we've
been
looking
that
we've
actually
sort
of
landed
on
calling
it
an
initial
value
with
to
try
to
convey
that
it's
a
value
that's
set
initially
when
an
instance
of
the
entity
is
created,
but
it's
then
not
used
anymore,
as
is
how
we've
been
thinking
about
that
from
our.
A
B
Sorry
on
the
gtl
side
or
or
on
dtdl,
so
we
haven't
so
constant.
We
actually
haven't
looked
at
constant.
Yet
I
was
just
talking
about
the
the
idea
of
default
and
we
landed
with
initial
because
we
wanted
to
convey
the
meaning
that
if,
if
you
had
an
entity-
and
it
got
an
initial
value,
but
then
you
ch,
you
then
removed
the
value.
It
would
go
back
to
the
default
value.
It
would
go
back
to
having
no
value,
and
so
we
have
we.
We
actually
have
that
semantic
of
it.
B
F
D
A
Good
point
I
mean
we're
very,
very
good
discussions,
but
but
yeah
it's
rupees,
maybe
to
wrap
up
this
segment
soon.
So
we
have
time
for
the
remaining
ones.
So
yeah
I
mean
we
already
had
a
good
point
on
the
on
the
way
forward
cars-
and
I
were
crafting
a
few
more
things
here
that
we
also
some
of
them
discussed
already,
but
if
go
quickly
through
the
money.
A
First
of
all,
requests
would
be
like
who
is
doing
these
t3
models,
and
could
we
have
something
like
a
call
for
action
to
contribute
those
models
because,
as
we
discussed
earlier,
it
would
be
great
to
have
more
ddl
models
to
work
on
this
conversion
see
how
we
can
how
we
can
convert
things,
but
then
also
in
the
1dm
activity.
We
are
collecting
models
from
from
different
ecosystems
and
would
be
awesome
to
have,
of
course,
also
ddl
models
out
of
that.
A
B
F
B
It's
it's
yeah,
it's
iot,
plug-and-play,
and
so
what
we
have
for
that
is.
We
have
a
model
repository
where
device
manufacturers
can
drop
their
models
in,
and
it's
a
public,
github
repo,
basically
and
and
so
I'll
drop
a
link
in
to
the
chat.
So
it's
a
great
place
to
go
see.
You
know,
example
models
that
actually
you
know
third
parties
have
done
and
are
have
published
about
their
devices.
B
That's
a
good
question
on
the
individual
models
I'll
have
to
check
on
what
we
do
about
that
on
those
yeah,
because
I'm
not
seeing
for
any
given
individual
one
that
we've
got
a
license
file
there
there's
a
license
for
the
whole
github
repository,
but
I
suspect
that
might
be
there
might
be
a
different
licensing
here.
So
let
me
check
on
that
and
follow
up
on
that.
A
F
Editorial
comment
that
when
we
had
our
semantic
workshop
in
2015,
was
it
that
the
microsoft
folks
there
proposed
an
approach
where
they
had
people
contribute
their
own
models
and
didn't
try
to
align
on
a
common
meta
model.
And
I
think
this
is
this.
Is
we
can
see
how
that
worked
out
versus
what
we're
doing?
And
we
can
maybe
work
out
a
single
way
of
sharing
these
two.
A
Okay,
excellent,
then
we
already
did
discuss
this
evolving
sdf
to
improve
support
for
ddl
model,
so
at
least
this
relations
is
something
what
we
are
already
already
adding
to
sdf.
Then,
let's
see
also
if
there
are
other
things
that
we
should
add
there,
or
also
taking
into
account
the
evolution
of
dddl.
A
We
already
briefly
talked
about
the
item:
idf
hackathon
activity.
So
of
course
it
would
be
great.
I
mean
if
some
of
you
are
in
in
the
team
are
able
to
join
that
and
maybe
think
about
how
we
can
improve
the
conversion.
Capabilities
between
stf
and
ietf
and
carson
will
be
sending
more
information
about
that.
I
had
an
activity
on
the
rg
list
and
then
you
can
see
if
that
perhaps
fits
your
schedules
and
you
have
possibilities
to
contribute
there.
B
A
Okay,
let's
do
that
and
then
also
I
mean
like
at
first
one
of
my
colleagues
was
looking
at
okay.
How
can
I,
how
can
they
visualize
stf
models
and
hey?
We
could
actually
use
some
of
the
azure
dtdl
tools
to
visualize
this
model,
so
if
you
just
convert
them
to
dtdl,
they
are
that
kind
of
interesting
potential
on
also
you
know
sharing
tools
across
ecosystems.
A
C
So
sorry,
I'm
late-
this
is
michael
mccool.
Sebastian
cabbage
should
be
joining
us
soon.
So,
in
addition
to
inner
conversion
with
sdf,
we
also
like
to
look
at
under
conversion
between
what
what,
with
things
theme,
models
and
thing
descriptions
it
might
be
via
sdf,
and
so
it
may
be
that
what
we
want
to
do
here
is
conversion
a
common
set
of
features,
and
maybe
we
can't
and
semantics.
C
C
A
A
C
Day
this
all
gets
dumped
into
an
rdf
database,
and
then
you
then
derive
new
information
from
it.
So
if,
if
we
all
agree
that
to
use
a
rdf
and
just
nlv
and
this
sort
of
thing,
it
makes
life
a
lot
easier
to
put
the
information
all
in
the
same
place
and
anything
informational
or
slightly
different,
at
least
if
we
could
agree
on
some
foundational
technologies.
That
would
be
helpful.
C
A
Good
okay,
but
I
think
I
mean
overall,
we
have
a
set
of
things
we
could
be
looking
together,
I
mean
now
now
summer
vacations
are
starting
in
europe
pretty
soon,
and
people
will
be
gone
until
until
august,
except
for
the
ietf
in
in
between,
but
maybe
you
could
think
about
something,
maybe
september
time
frame.
A
follow-up
on
on
this
evdl
discussion.
A
Thanks:
okay,
so
then
yeah,
let's,
let's
touch
base.
You
know
when
we
are.
I
mean
at
the
hackathon
at
the
latest
when
we
are
all
back
from
vacations
and
and
plan
for
that
and,
of
course,
there's
also
potential
for
those
who
don't
take
the
long
vacation
to
progress.
Some
of
these
work
items
very
good.
Okay,
thanks
everyone.
I
know
we're
a
bit
bit
overtime
already,
but
milan.
How
do
you
think?
A
Can
we
well?
First
of
all
I
mean
we
do
have
our
our
third
our
agenda
item
and
we
were
discussing
that.
Okay,
we'll
use
how
much
how
much
time
we
have
left
for
that
and
now
it
seems
we
went
unfortunately
over
time
with
the
digital
discussion.
So
what
I
would
suggest
those
of
you,
of
course,
who
have
to
leave
half
past
we're
very
welcome
to
do
so,
but
if
some
of
our
others
have
time
to
go,
let's
say
10
minutes
over
time.
A
G
Yeah-
and
I
mean
it
would
be
actually,
as
I
mentioned,
the
email
to
cool
to
have
would
be
cool
to
have
more
time,
because
there
are
some
great
news
in
the
thing
model
which
you
can
do
for
modeling
and
and
to
create
thing
description
out
of
that,
and
we
have
also
some
experience
with
sdf
definition
and
to
have
this
as
a
thing
model
representation.
I
think
to
be
realistic.
C
C
C
G
And
since
since
we
are,
we
have
all
here
and
around
maybe
just
off
topic
here,
but
we
have
this
open
day
and
and
maybe
it
would
be
also
cool
to
have
from
your
side
the
latest
from
sdf
and
and
brian
you're.
Also
here,
maybe
it
would
be
kind
of
cool
to
have
also
the
latest
from
the
detail,
dr
language.
Maybe
you
have
also
time
to
give
us
a
presentation
on
that
and
and
michael
mike
mccoo,
when
when
is
planned
open
day,
so
maybe
we
can
already
share
the
date
here.
It's
the
28th.
C
A
H
So
this
is,
by
my
account,
the
third
update
on
the
topic,
and
this
time
there
are
some
there's
some
technical
progress
to
report.
So
just
as
a
quick
reminder,
this
is
basically
the
deep
tier
trg
paper
that
describes
key
features
of
iot
data
standards
that
we've
been
talking
about
for
a
while
data
standards
referring
to
information
models,
not
all
of
them
that
he
described
them
in
a
consistent
manner.
H
One
of
my
goals
is-
and
I
don't
know
if
there
is
a
group
agreement-
we
haven't
discussed
it
as
a
consensus-
is
that
such
a
document
should
be
usable
by
researchers
and
iot
practitioners,
of
course,
for
selection
and
use
of
things,
but
not
just
standards,
definition,
organization,
warms
and
insiders.
In
other
words,
not
just
the
fine
points
of
comparison
of
standards,
but
rather
helping
somebody
who
is
outside
to
understand
the
lay
of
the
land
and
where
various
things
fit
and
how
they
may
fit.
H
So
in
terms
of
startups
update,
I
have
written
in
mid-april
the
first
draft
of
the
thing
and
I'll
describe
briefly.
What's
in
it
and
at
this
point
it's
it's
written
in
the
internet,
rfc
product
drafts
format,
one
of
my
favorite
publishing
models,
and
it
is
essentially
a
description
of
the
description.
Basically,
what
should
be
the
standard
description
and
outlining?
H
Essentially
what
and
why
so?
Currently,
the
content
of
the
document
is
is
basically
this
major
bullets
are
the
sections.
So
there
is
the
introduction
and
problem
statement.
You
know
what
is
infection
modeling?
Why
do
you
need
semantic
interoperability
in
iot?
That
is
machine
to
machine
communication
that
has
to
be
machine,
interpretable,
meaning
of
data
and
goes
into
something
tell
what
the
meaning
of
data
in
this
context
mean,
basically
that
the
machines
have
the
content.
H
They
can
recognize
a
particular
type,
an
object
and
know
how
to
process
it,
assuming
that
they
are
so
equipped,
and
then
the
next
chapter
does
the
next
section.
The
brief
overview
of
iot
information
models
mostly
object.
What's
the
structure,
object,
types,
properties,
attributes
interactions
and
links
and
iot
frameworks?
Basically,
is
a.
H
It
could
be
serialization
protocol,
binding
security
management
or
some
combination
of
those.
But
the
main
point
is
in
order
to
comply
and
use
the
framework.
One
has
to
implement
the
entire
framework
and
be
compatible
with
it
in
addition
to
the
information
model
itself.
So
that's
basically
introductory
and
sort
of
orientational
discussion
up
to
this
point,
and
then
there
is
an
outline
of
what
should
be
in
the
common
criteria.
What
should
be
covered
in
roughly
in
in
what
way,
and
currently
the
sections
of
objectives
and
purpose
basically
state.
H
One
thing
that
I
haven't
added-
and
probably
it
could
be
just
an
elaboration
in
this
section-
is
to
have
what
is
the
basic
conceptual
approach
to
the
problem
of
that
particular
standard
and
what
are
sort
of
the
key,
architectural
and
design
principles
sort
of
the
design
foundation
that
is
based
on
then.
The
next
thing,
which
is
important,
I
think,
is
the
domain
of
usage.
If
there
is
a
primary
target
vertical
verticals,
it
would
be
nice
to
state
that
usually
many
of
the
standards
that
I've
seen
you
say
are
they
applicable
to
all
things
iot.
H
H
That's
a
within
the
description,
the
other,
I
think
important
thing
is
the
environmental
assumptions.
Basically,
what
is
assault
to
be
in
place
for
the
specification
to
work?
In
other
words,
what
does
it
provide
and
define
and
what
else
it
assumes
or
doesn't
define
but
needs
to
be
there
in
order
to
work
so
that,
because
sometimes
that's
left,
often
as
an
exercise
to
the
reader.
H
H
H
Is
it
something
like
crowdsourcing
that
ocf
does
and
hasten,
or
do
you
have
to
wait
for
the
high
priest
to
define
the
object
before
you
can
use
it,
and
then
can
you
bring
forward
a
proposal?
That's
basically
that
that
is
sort
of
architecturally
compliant
with
with
the
principles
and
design
practices.
C
A
C
H
C
H
H
H
Often
you
know
to
be
derived
by
osmosis
in
terms
of
what
the
node
is
supposed
to
be
created
to
be
able
to
do
at
the
design
time.
In
other
words,
what
the
code
is
looking
for
right
for
some
basic
properties
of
things
versus
runtime.
What
can
it
expect
to
discover
and
how
to
interpret
those?
Because
when
you're,
you
know
designing
writing
the
code
for
the
node,
you
need
to
know
those
things
right.
So.
H
C
H
I
agree,
and
web
of
things
is
one
of
the
few
who
separates
that
nicely
and
you
know
fairly
clearly
in
the
sense
from
the
design
point
of
view
and
something
like
that
for
the
others
at
least,
to
indicate,
what's
assumed
right
right
right
and
if
there
are
tools
and
reference
implementation,
especially
for
frameworks
that
helps
a
lot,
because
you
have
to
implement
the
framework
it's
much
easier.
If
you
have
something
to
start
with
and
then
the
licensing
terms
are
also
good
to
know.
H
You
know
whether
you
have
to
be
a
member
of
the
club
or
you
have
to
pay
the
feel
you
just
get
it,
and
so
these
were
basically
the
things
sort
of
that.
I
have,
as
I
said,
put
in
this
document,
and
and
we
have
it
in
in
on
github,
but
I
think
it's
a
closed
repository
carsten
has
access
and
the
two
michaels,
but
so
the
the
next
steps
are,
you
know,
okay,
so
we
need
to
elicit.
H
So
I
would
like
to
elicit
the
workroom
feedback
and
contributions,
and
you
know,
revise
the
document
accordingly,
so
we
may
make
it
perhaps
more
visible
and
there
are
at
least
two
types
of
comments
that
you
know.
All
comments
are
welcome,
but
there
is
the
lights
touch,
which
would
be
greatly
appreciated
and
it
doesn't
take
much
time
is
basically
comments,
sort
of
high-level
comments
on
the
direction
yeah.
We
need
this,
so
we
don't
need
this
and
things
to
include
or
exclude.
H
You
know
have
these
things
you
have
to
have
this
and
some
other
things
that
may
be
less
of
less
interest,
and
you
know
I
have
my
opinion,
but
that's
my
opinion.
There's
also,
you
know
it
should
be
a
group
consensus
in
order
to
be
useful
and,
of
course,
contributions
like
you
know.
If
somebody
wants
to
write
a
section
yet
and
be
added
to
the
list
of
us,
that's
great
and
then
the
next
thing
is
to
produce
the
our
work
group
approve
the
document.
Sorry,
I'm
having
a
problem.
H
To
produce
the
approved
document,
I
don't
know
that
there
is
a
process
for
proving
and
then
potential
future
expansion
is
to
work
with
sdos
and
interested
stu
if
they
want
to
contribute
to
the
description
using
this
format
who
work
collaboratively
in
accomplishing
that,
and
I
can
see
two
coins
of
that.
You
know
it
would
be
great.
D
Thank
you
so,
with
respect
to
the
process,
the
indeed
is
a
process
that
usually
goes
through
a
number
of
stages,
so
we
we
will
have
enough
stages
that
we
can
collect
the
comments
from
from
various
people
and
of
course
the
first
stage
is
getting
an
internet
draft
out
there
and
the
the
getting
comments
on
that
and
so
on.
And
then
at
some
point
we
are
going
to
make
a
research
group
adoption
call
which
asks
the
research
group
whether
the
research
group
believes
this
is
useful
to
work
and
wants
to
commit
time
on
on
contributing.
D
But
the
point
about
in
rfc
is
that
it
may
be
called
request
for
comments,
but
this
is
really
the
the
point
where
we
actually
no
longer
actively
looking
for
comments
where
we
we
are
always
actively.
Looking
for
comments,
but
we
publish
things
as
rfc,
where
we
think
it's
it's
already
stable
enough,
that
people
will
be
able
to
live
with
it
for
a
year
or
two.
A
D
H
D
I
actually
must
admit
I
haven't
seen
a
version
of
this
document
for
two
months
now,
so
I
think
we
we
maybe
should
look
at
this
once
more
and
then
publish
it
as
an
internet
draft,
because
I
mean
the
word
publish
really
has
very
many
meanings
in
the
ietf,
so
this
means
it's
out
there.
We
can
ask
people
to
to
comments,
make
comments
and
and
people
see
it
and
then,
after
a
few
more
rounds
which
can
happen
very
quickly.
D
H
I
was
to
appeal
to
the
two
michaels
who
you
know
have
a
lot
of
stuff
in
the
tables
written
in
this
space
to
at
least
send
me
the
stuff,
the
raw
stuff
that
they
may
have,
that
that
is
kind
of
relevant
to
this,
and
then
I
can,
if
necessary,
I
can
even
try
to
fit
them.
You
know
in
the
context,
without
them
having
to
do.
C
Yeah,
well,
I
I
just
updated
my
repo
to
catch
up
with
what
the
latest
version
is,
and
I
think
I
need
to
do
is
post
an
issue
to
link
to
the
stuff,
because
some
of
it's
kind
of
big
but
I'll,
see
where
I
can
stick
it
in
the
repo
and
I'll
have
some,
maybe
powerpoint
initially,
but
we'll
I'll.
Do
that
at
least
and
then
we'll
go
from
there.
C
D
D
Good,
it
seems
we
have
reached
the
end
of
our
time.
So
sebastian
probably
was
right
that
we
should
postpone
this,
so
I'm
really
interested
in
hearing
the
the
update.
So
I
think
we
should
advertise
this
this
open
day,
so
people
in
the
research
group
and
people
in
wishy
know
about
that
find
a
time
in
in
september,
in
early
september,
where
we
can
reconvene
because
most
of
europe
is
going
into
various
forms
of
shutdown
right
now
for
once
not
covered
shut
down,
but
vacation
is
down.
A
C
D
Well,
I
would
just
like
to
point
out:
we
do
have
the
the
hackathon
in
in
the
week
before
itf
111.
So
if
you
have
the
ability
to
to
join
there,
please
tell
me
and
please
think
about
what
what
kind
of
converters
and
other
processors
you
want
to
bring
to
the
hackathon
and
maybe
also
play
with
the
sdf
yang
converter,
that
became
accessible
today.
So
if
you
have
an
sdf
model
that
you
wanted
to
see
how
it
would
look
like
in
yang,
please
try
that
and
please
try
the
inverse
as
well.