►
From YouTube: T2TRG Interim Meeting, 2021-03-02
Description
T2TRG Interim Meeting, 2021-03-02
A
B
Yes,
thank
you.
So
we
have
been
running
the
the
hackathon
for
about
24
hours
now
and
apparently
very
few
people
actually
had
lots
of
time
to
prepare
something.
B
So
what
what
we
said?
What
we
wanted
to
look
at
in
this
hackathon
is
essentially
five
things.
Actually
we
first
said
four,
and
then
we
found
we
have
to
do
the
fifth
as
well
validating
sdf
1.1,
which
is
in
in
a
consensus,
call
in
the
asdf
working
group
right
now,
so
just
making
sure
that
our
all
of
our
tools
and
all
of
our
models
are
up
to
sdf
1.1.
B
The
second
thing
is
to
continue
work
on
the
semantic
proxy
idea,
so
that
that
has
been
around
for
a
while
and
we
still
plan
to
have
oma
and
ocf
implementations
talk
to
each
other,
guided
by
common
sdf.
So
no
no
ad
hog
support
for
the
other
ecosystem
in
in
the
respective
implementations,
but
trying
to
glue
them
together
by
a
common,
sdf
and
mapping
files,
and
that's
really
the
the
focus
of
this
hackathon.
B
From
my
point
of
view,
getting
this
mapping
file
concept
a
little
bit
further
off
of
the
ground,
so
we
we
all
have
done
something
in
the
space
of
mapping
files,
but
we
at
some
point.
We
probably
need
to
turn
this
into
a
coherent
vision
and
another
thing
that
we
want
to
do
is
we
want
to
spend
work
specifically
talking
to
the
web
of
things
ecosystem
so
that
there
is.
B
There
are
thing,
descriptions
that
have
a
lot
of
things
that
that
sdf
doesn't
have
so
is
there
something
that
that
would
go
into
sdf
or
is
there
something
that
would
go
into
mapping
files
and
so
on?
So
that's
something
you
want
to
look
at
and
finally,
it
turns
out
that
we
probably
need
a
little
bit
stronger
consensus
on
the
processing
model.
So
how
do
we
actually
make
use
of
all
this
for
a
single
sdf
model?
It's
pretty
clear
what
this
does.
B
It
describes
the
the
interactions
and
and
the
the
data
that
that
is
going
into
and
coming
out
of
the
interactions
of
of
a
specific
aspect
of
a
thing,
but
what
does
it
mean
to
to
have
a
mapping
file
so
how?
How
does
the
mapping
file
interact
with
one
or
more
existing
models,
and
and
how
do
you
actually
cross
link?
B
How?
How
do
you
do
references?
How
does
the
name
spacing
work
and
so
on?
So
this
is
a
rough
overview
of
what
we
were
trying
to
do
and
before
I
go
into
more
detail
just
on
the
mapping
file
concept,
just
a
little
bit
more
about
the
the
logistics.
B
So
we
want
to
have
a
daily
coordination
call,
and
this
time
it
will
happen
right
after
the
wishy
call.
We
had
one
on
monday
and
we
plan
to
have
one
on
wednesday,
so
these
dates
were
pretty
much
determined
by
other
meetings
and
on
thursday
we
still
have
a
choice
and,
and
probably
we
will
move
to
1600
on
thursday.
But
let's
see
we
just
have
to
decide
that,
so
we
do
this
in
a
relax
room
that
that
is
permanently
open
during
the
the
entire
hackathon.
B
And
if
you
want
to
click
on
these
links
you
you
cannot
do
this
in
in
webex,
but
of
course
these
slides
are
in
the
wishy
github
repository.
So
if
you
go
to
a
github,
trg
wishy,
then
you
can
find
this
these
slides
under
slides
and
can
click
on
things.
So
there
is
a
link
to
the
wiki.
There
is
a
link
to
the
day-to-day
notes
that
you
can
look
at.
B
Okay,
so
back
to
to
mapping
files,
the
the
mapping
files
are
a
general
concept
of
of
doing
something
doing
some
some
specification
or
modeling
work
that
somehow
attaches
to
sdf
specifications
and
bridges
these
specifications
into
something
and
that
something
can
be
a
whole
ecosystem.
So
there
might
be
an
ipso
mapping
file,
for
instance,
that
would
be
used
in
an
oma
environment.
B
So
for
ipso
the
the
most
important
thing
that
people
have
been
talking
about
is
assigning
ids
because
in
in,
if
so,
everything
has
numbers,
and
we
we
didn't-
want
to
put
these
numbers
into
the
common
sdf
models,
because
these
numbers
only
really
mean
something
in
the
ipso.
The
I
a
lightweight
mtm
environment,
so
being
able
to
write
a
mapping
file
that
that
supplies
these
ids
to
generic
scf
models,
sound
like
a
good
idea,
and
we
we
have
examples
for
that
in
the
repositories.
B
But
we
also
have
an
example
in
the
exploratory
repository
that
that
goes
a
little
beyond
that
and
actually
not
just
defines
ids
but
also
defines
representation
data
types.
So
the
the
data
types
in
the
sdf
file
are
on
the
information
model
level,
but
at
least
that's
what
we
are
trying
to
do
here
and
we
may
need
to
transform
these
into
ecosystem
specific
types
like
16
bit,
values
or
something
which
maybe
are
not
meaningful
common
type.
B
So
we
don't
want
to
do
this
as
the
the
sdf
level,
but
the
the
mapping
file
should
be
able
to
provide
the
information.
I'll
have
an
example
for
that
in
a
moment.
So
that's
one
category
of
mapping,
files,
ecosystem,
specific
mapping
files,
so
they
would
apply
to
any
use
of
an
sdf
model
in
an
ecosystem.
So
you
would
have
a
pretty
good
idea
whether
you
are
applying
this
mapping
file
at
a
specific
time
or
not,
because
you
would
know
which
ecosystem
you
are
in
now.
B
B
So
these
are
really
things
that
are
specific
to
an
instance,
and
they
may
also
be
non-technical
instance
information.
So
you
might
want
to
know
whether
the
the
the
temperature
sensor
you
are
talking
about
is
is
out
on
on
the
terrace
or
is,
is
inside
the
living
room,
or
is
it
the
temperature
sensor
that
that
measures
the
room
temperature
or
is
it
the
temperature
sensor
that
measures
the
inside
temperature
of
of
the
fridge?
B
And
finally,
we
probably
need
data
models
for
these
instant
mappings,
so
the
the
protocol
parameters
these
are
in
in
some
instant,
specific
mapping
files.
But
then
you
want
to
make
sure
that
the
instance
specific
mapping
files
actually
make
sense,
because
you
have
many
of
them.
You
don't
have
a
single
one
that
is
carefully
crafted
by
by
by
a
standard
maker.
You
you
have
many
of
them,
so
you
essentially
need
a
data
model
for
these
instant
instance
mapping
files.
B
So
that's
probably
one
of
the
fun
parts
that
that
we
have
to
look
at
and,
as
I
mentioned,
we,
we
need
to
have
some
some
idea
of
how
an
implementation
would
find
out
which
of
these
many
instant,
specific
mapping
files
to
actually
activate
so
how?
How
do
these
mapping
files
actually
come
into
force?
B
If
you
have
20
of
them,
which
one
are
you
using,
so
let's
talk
about
these,
this
mapping
file
schema
thing
that
I
just
alluded
to
so
sdf
has
been
terminally
confusing
to
people
because
we
have
two
levels
here
we
have
the
the
sdf
model,
which
is.
B
Represented
in
json
and
we
have
a
formal
description
technique
and
we
are
using
both
facility
ellen
jason,
schema
org
here,
that
is
in
the
sdf
internet
draft
that
describes
how
these
models
are
looking
like.
So
we
have
a
meta
model
that
describes
the
model
and
the
the
model
itself
talks
about
data
and
just
to
confuse
things.
It
uses
data
schemas
that
look
a
lot
like
json
schema
org.
B
So
when
people
have
been
talking
about
the
schema
with
respect
to
to
sdf,
you
never
knew
whether
they
were
talking
about
the
the
meta
model
or
the
model,
and
now
it's
getting
worse,
because
we
not
only
have
the
the
model
and
the
the
which
has
a
specific
structure
and
some
data
schemas
that
are
in
that
model.
But
we
also
have
a
layer
three
or
level
three,
which
would
be
a
meta-meta
model
for
the
meta
models
that
specify
what's
in
in
a
mapping
file.
B
So
that
will
require
a
lot
of
discipline
in
in
discussion
to
to
point
out
which
which
of
them
are
we
talking
about
so
one
one
of
the
layer.
Three
models
that
people
have
been
using
so
far
is
a
simple
one
of
augmentation,
so
you
take
an
sdf
specification
and
you
just
augment
the
the
qualities
that
are
in
the
sdf
specification
by
additional
ones
and
and
you
can
put
any
quality
there
you
want.
B
So
the
the
other
thing
that
came
up
is
that
when
we
model
something
in
a
mapping
fire,
we
may
not
just
be
modeling
things
by
by
by
adding
information
to
the
sdf
file.
But
this
what
we
are
actually
modeling.
There
is
a
transformation.
B
B
Yes
and
yeah,
I
was
not
really
trying
to
to
get
the
first
bullet
crystal
clear.
I
think
we
have
to
still
have
to
discuss
that.
The
second
bullet
is
about
mapping
actually
mapping,
so
the
the
abstract
sdf
model
may
talk
about
the
temperature
and
in
the
abstract
model,
for
instance,
we
might
decide
well,
okay.
B
So
we
will
need
to
do
a
linear
transformation
here
and
obviously
not
all
transformations
are
just
linear.
So
there
are
a
million
color
models,
for
instance,
which
are
probably
the
the
worst
example
of
trans
non-linear
transformations,
but
it
may
also
be
something
simpler
like
going
from
from
a
power
or
a
voltage
to
a
level.
So
you,
you
have
a
logarithmic
relationship
and
you
may
even
need
to
do
structural
transforms
because
somebody
is
describing
a
light
by
by
its
intensity
and
it's
the
color
temperature
and
somebody
else
is
describing
it
by
rgb
values.
B
So
I'm
not
saying
that
that
we
need
to
be
able
to
express
all
that,
but
I'm
saying
we
maybe
need
to
be
in
a
position
to
actually
point
to
places
where
these
are
expressed,
so
that
that's
where
the
transformation
comes
in,
it's
not
sufficient
to
say
this
is
an
rgb
value
and
and
then
I
have
a
common
model
that
says
it's
a
whatever
an
lab
value,
but
you
need
to
to
explain
how
the
transformation
actually
happens.
B
A
Quick
comment:
I
mean
on
the
some
of
the
basic
units
thanks
the
cinema
registry.
We
actually
have
the
linear
transformations
covered
to
some
extent
programmatically.
B
Yeah
as
long
as
you're
talking
about
units,
I
think
we
are
indeed
covered.
But
then
there
are
things
like.
Is
this
an
8-bit
value
or
a
16-bit
value,
and
are
we
using
an
offset
representation
or
two's
complement
representation
and
so
on
and
so
on?
So
there
are
lots
of
things
that
are
going
on
when
you
go
in
into
device
specific
data,
and
we
need
to
be
able
to
talk
about
that.
B
Okay,
so
let's
just
finish
this,
I'm
already
five
minutes.
I
was
with
you
with
looking
at
a
few
examples,
so
this
is
the
the
basic
id
thing,
so
we
we
have
an
sdf
object
level
and
this
gets
an
id
h.
So
this
is
not
an
ipso
example,
but
the
epso
example
is
look
similar.
B
That's
simple,
augmentation
great.
The
the
only
thing
that
that's
missing
for
me
is.
I
would
like
to
be
able
to
say:
ipso
ids
are
between
1
000
and
10
000,
or
something
like
that.
I
currently
cannot
say
that
I
can
write
the
the
mapping
files,
but
I
cannot
help
yeah
check
these
mapping
files.
So
that's
one
place
where
we
we
don't
have
the
meta
model
that
would
metameter
model.
That
would
help
us
and
the
other
observation
is.
B
This
kind
of
mapping
would
always
be
active,
so
you,
you
know
that
in
a
zcl
environment
you
would
have
this.
This
mapping
active,
so
we
essentially
just
need
some
something
in
the
info
block
of
this
mapping
file
that
tells
us
what
what
area
of
application
this
has
so
that
that's
relatively
easy
now.
B
One
other
thing
that
michael
had
in
in
his
example,
was
mapping
the
the
abstract
values
in
the
sdf
specifications
to
to
concrete
values
for
the
ecosystem,
so
in
in
this
ecosystem,
move
step
mode,
values
have
a
zero
for
up
and
one
for
down,
and
so
we
are
essentially
using
a
json
schema
org
a
schema
here
to
to
provide
a
constant
value-
maybe
that's
overkill,
but
at
least
something
we
all
understand
at
this
point
in
time
and
yeah.
B
This
works
great
for
sdf
choice,
because
we
have
a
discrete
addressable
values
here.
Each
value
that
is
possible
has
a
json
pointer,
so
we
can
establish
a
mapping
nicely,
but
as
soon
as
we
have
larger
value
spaces,
16-bit
32-bit
floating
point
whatever
that
doesn't
really
work
anymore
and
again,
yeah
so
maybe
being
in
in
the
zigbee
environment,
sigma
cluster
library,
environment
activates
this
automatically,
but
yeah.
B
That
becomes
a
little
bit
more
interesting
here
so
on
on
this
slide
here
we
could
simply
not
use
id
but
use
zcl
id,
and
we
would
have
a
clear
separation
from
an
ipso
id
that
might
also
be
be
assigned
here.
On
this
slide,
it's
actually
the
data
video,
so
that's
not
quite
clear
whether
we
have
a
zigbee,
zero
and
and
oma0,
and
I
think,
there's
only
one
zero.
B
So
we
need
a
different
way
to
actually
activate
this
mapping.
B
So
how
does
the
the
transformation
info
work?
So
that's
from
the
example.
The
example
currently
just
says
that
that's
16
bits,
so
we
have
2
to
the
16
values,
but
it
doesn't
tell
us
how
we
are
going
to
use
these
bits.
So
are
we
tools
complement
or
offset,
or
there
are
many
ways
to
to
skin
this
cat,
and
then
we
have
the
the
instance
specific
information
and
actually
not
all
of
this
information
is
instant
specific.
B
So
it's
going
to
be
all
just
meshed
up,
so
we
we
are
simply
using
a
web
of
things.
Protocol
binding
here
with
an
href
and
and
then
some
modbus
specific
things
like
unit
ids
and
entities
and
offset
and
length
and
so
on.
So
these
are
register
numbers.
I
if
I
understand
this
correctly
and
interestingly
this
actually
this
example
shows
one
concept
that
we
probably
need.
B
The
the
href
actually
is
a
template.
It's
it's
not
a
final
value
because
because
it
says
this
ip
address,
so
when
we
go
to
an
actual
instance
mapping,
then
we
can
put
the
actual
ip
address
in
there.
But
if
we
are
talking
about
modbus
mapping
in
in
at
the
class
level,
we
need
to
say:
where
does
this
parameter
actually
come
from.
C
C
Well,
they
don't
use
the
complicated
subset.
Mostly,
we
just
use
the
simple
substitution,
but
I
should
also
mention
we
have
two
places.
We
use
them
both
the
model
level
and
at
the
and
there's
two
different
places.
We
need
them,
and
so
we
have
to
use
both
single
curly
braces
and
double
curly
braces
in
different
places
to
distinguish
the
two
cases.
B
Yeah,
I
think
we
that
that's
really
a
problem
that
we
need
to
understand.
So
one
thing
that
I
learned
from
the
html
world
is
that
when
sgl
html
was
used
for
more
and
more
complicated
applications,
the
html
people
tried
to
solve
problems
by
doing
syntactic.
B
Expansion,
so
so,
essentially
a
macro
level
style
thing,
and
this
this
made
some
things
possible,
but
it
really
made
it
hard
to
to
process
these
macros
in
any
meaningful
way,
so
they
could
only
be
created
by
humans
and
they
could
never
be
understood
by
by
machines
that
were
trying
to
do
anything
but
just
simply
applying
them.
So
I
think
we
need
to
to
discuss
at
some
point
whether
this
is
just
a
syntactical
thing
like
in
url
templates
or
maybe
a
little
more.
B
So
I
think
we
all
understand
something
needs
to
be
done
here.
So
there
is
something
like
a
concept
of
a
parameter,
so
jan
has
actually
provided
one
one
more
example
for
of
a
mapping
specification,
which
has
the
interesting
property
that
it
actually
doesn't
start
from
the
the
sdf
namespace.
But
it
comes
from
the
ecosystem
namespace.
B
So
there
is
a
co-op
server
here
and
that
exposes
two
paths
to
your
eyes:
switch
value
and
switch
on,
and
essentially
the
mapping
file
tells
you
from
there
which
sdf
object
and
sdf
property
actually
is
being
described
and
add
security
and
and
whatever
the
method
that
you
use
so
that
that's
a
different
way
of
looking
at
mapping
files.
C
C
I
was
going
to
say,
I
think,
what
tds
and
swagger
kind
of
transposes
of
each
other
in
exactly
this
nature.
One
has
a
list
of
interaction
like
urls
and
then
describes
the
interaction
and
the
other
has
a
list
of
interactions
and
describes
the
urls
and
they're
equivalent.
But
it's
a
little
confusing
when
you're
reading
one
and
expecting
the
other
ones.
B
B
To
give
a
brain
dump
on
some
more
things,
we're
doing,
and
what
I'll
let
you
okay.
So
let
me
finish
my
slides,
because
this
is
my
last
slide,
so
there
are
a
few
things
that
need
to
be
done
and
one
of
which
is
understanding
the
processing
model
better.
B
I
had
this
that,
on
the
the
first
slide
in
this
segment,
build
more
strawman
examples
like
the
above
and
and
use
these
strawman
examples
to
understand
the
elements
of
the
meta
and
metamata
models,
so
that
would
be
my
program
for
for
the
rest
of
the
hackathon
to
get
a
little
bit
further
in
in
these
three
items.
C
Okay,
so
I
just
want
to
say
a
few
things
about
what
so,
first
of
all
we're
working
on
a
couple,
things
that
are
related
to
things
you
mentioned.
C
C
Can
we
use
one
dm
and
sdf
source
semantic
annotations
in
thing
descriptions
and
I
think
generally,
we
want
to
do
as
much
as
we
can
to
make
these
things
consistent,
so
we're
not
using
different
syntaxes
for
the
same
thing
and
that's
the
thing
there's
one
more
part
I
want
to
mention
is
lately
we're
working
on
geolocation
information
for
tds,
and
so
I
have
a
proposal
and
I'll
just
share
a
link
here
in
a
new
minutes
me
a
second
to
find
it.
D
C
C
So
hold
on
right
right.
So
actually
decent
point
is
another
issue.
Let
me
get
back
to
that.
So
I
just
clicked
a
link
to
an
issue
where
we're
putting
at
the
other
geolocation
information
model
and
we're
talking
with
various
people
in
like
ogc,
and
you
know
geosparkles
like
this
trying
to
get
some
consistent
information
model
and
query
language
for
our
directories
to
be
able
to
query
locations
of
devices.
C
So
it'd
be
really
great
if
we
could
maybe
also
work
on,
for
example,
an
sdf
capability
for
geolocation
that
use
the
same
information
model,
because
if
the
information
model's
the
same,
we
can
interconvert
things,
but
it's
different
than
more
problem.
So,
let's
not
reinvent
the
wheel
here.
Let's
try
and
converge
on
that.
C
Regarding
json
pointers,
we've
recently
discovered
several
uses
for
json
pointers
for
referring
to
parts
of
tds.
So,
for
example,
geolocation.
C
You
know
we
could
have
static
a
dynamic
view,
location
so
is
location
embedded
in
the
td
or
is
it
in
a
property?
And
so
the
idea
is,
we
add,
a
link
which
has
a
json
pointer,
which
then
points
to
where
the
geolocation
is.
It
might
be
in
the
same
td,
it
might
be
in
a
different
td,
it
might
be
in
a
property
and
a
thing
and
the
same
thing
or
a
different
thing,
and
so
the
idea
is
to
use
a
json
pointer
for
that.
C
C
I
think
we
should.
You
know
converge
on
a
common
syntax
for
that
and
finally,
as
I
just
mentioned
before,
tms
actually
include
uri
templates
and
we
also
have
uri
templates
for
data
models
embedded
in
things.
So
we
have
two
different
levels
of
parameters
and
we
actually
use
a
double
curly,
brace
and
thing
models
on
a
single,
curly,
brace
or
data
model
type
uri
parameters
to
distinguish
them
and
be
kind
of
nice
to
discuss
how
we
use
uri
templates.
You
know
and
maybe
come
to
some
consistent
agreement
on
what
parameters.
Look
like.
C
Those
are
a
few
things
and
I'm
going
to
be
attending
the
idf
hackathons.
I
can
only
attend
today's
and
tomorrow's
meetings
because
it's
conflicts,
so
what
I've
been
doing
is
we
can
create
issues.
I
just
pointed
to
two
of
them
in
the
watt
plug
fest,
which
is
happening
concurrently
and
you
know.
Hopefully
we
can
work
on
things
in
each
tracker
or
in
the
watt.
Confess
meetings.
C
Okay,
I'll,
stop
my
brain
dump
there.
But
if
people
have
questions,
we
can
certainly
discuss
things
more,
oh
and
also
finally,
one
more
thing
for
the
directories:
we
actually
have
three
query
languages,
so
we
have
sparkle,
okay
for
semantic
stuff.
We
have
a
json
path
and
one
thing
I
need
to
do
is
look
at
the
recent
json
path.
Standardization
activity
in
ietf,
one
of
our
problems
is
jsonpath,
is
not
a
standard,
it's
just
a
blog
post,
basically
and
then.
C
So
one
thing
we're
going
to
look
at
is
you
know,
ideally
there'd
be
some
json
path,
syntax
that
was
semantically
equivalent
to
the
xpath
syntax,
so
you
can
interconvert
them
because
they're
both
syntactic,
basically
as
opposed
to
semantic.
C
So
it's
another
activity
that
I
think,
if
you
guys
are
doing
the
directory
service
or
that
sports
queries
we
should
also
negotiate
on
and
actually
for.
Geolocation
is
also
going
to
be
a
set
of
queries
and
for
geolocation
it
turns
out.
We
just
identified
two
classes
of
queries,
one
is
point
and
radius,
so
like
latitude
and
longitude
accuracy,
the
other
is
actually
point
direction
and
field
of
view
which
shows
up
in
augmented
reality.
C
B
Okay,
so
I
used
exactly
double
the
amount
of
time
that
we
have
allocated,
so
maybe
we
should
quickly
jump
to
the
next
item,
which
is
this
technology
landscape
update.
A
A
A
B
But
we
do
see
your
cursor
very
nice.
A
A
Okay,
what
we
could
do
is
do
a
few
minutes
on
the
whispy
way
forward.
While
we
wait
for
milan
to
get
his
audio
working
so
belongs.
Let
us
know
when
you
got
things
working
and
then
we
can
jump
back
to
your
your
statement.
A
Okay,
so
so
this
is
very
freeform
segments,
as
I
mentioned
in
the
beginning
of
the
intro,
we've
been
very
much
focused
on
the
asdf
and
sdf-1
vm
work
in
in
the
in
the
wish
so
far,
and
we
certainly,
as
karsten
already
identified
this
bunch
of
things
we
can
be
working
on
on
sdf
or
also
also
going
forward.
A
A
One
aspect
that
is
very
much
related
to
the
astf
is
we
have
the
discussion
with
microsoft,
azure,
digital
twin
definition,
language
are
folks,
so
I
think
that's
something
that
we
should
pick
on
and
also
think
about
how
it
fits
to
the
asdf
work,
but
then
also
in
general,
of
on
our
quest
of
understanding,
the
landscape
better
and
then
the
whole
topic
of
other
hyper
media
and
semantics
topics.
A
E
Work
niklas
here,
I
I
think
it's
I
mean
bringing
in
the
microsoft
folks,
she
said
is
a
good
thing
and
if
we
can
do
the
same
with
other
ecosystems,
that's
also.
I
think
this
could
be
a
good
venue
for
a
good,
maybe
more
neutral
venue
than
asdf
directly
to
to
lift
these
things
in
disgusting.
D
I
think
the
similar.
Similarly,
I
think
that
there
are
a
couple
of
things
that
are
sort
of
gaps
and
missing
right
now.
One
one
of
them
is
when
you
look
at
the
kind
of
models
that
are
created
with
opc
ua
and,
to
some
extent
even
digital
twin
models.
D
They
include
modeling
a
lot
of
physical
concepts
that
aren't
really
iot
gadgets
or
don't
really
have
you
know
affordances
and
interfaces,
but
they're
still
necessary
to
make
the
the
whole
graph,
and
we
need
to
at
least
figure
out
how
to
connect
to
those
and
maybe
some
standard
ways
of
expanding
the
work
out
into
the
sort
of
I've
been
calling
it
feature
of
interest,
but
there's
really
a
lot
more
to
it
than
that.
I
think
there's
sort
of
are
quantities
and
units,
as
we
were
looking
at
earlier.
D
There
are
some
places
there
where
it
doesn't
have
much
bottom
right
now
you
can
look
at
it.
Whoa,
there's
a
big
problem
here,
there's
a
big
thing
to
be
solved.
The
other
thing
is,
I
think,
we've
been
kind
of
wandering
around,
assuming
that
the
the
ultimate
iot
system
is
where
all
the
things
are
autonomous
and
talk
to
each
other,
and
there
are
no
intermediaries,
bridges
or
gateways
or
any
of
that,
and
I
think
that
that's
just
wrong.
D
I
think
we
need
to
acknowledge
that
the
systems
are
layered
and
think
about
start
thinking
about
that
a
little
bit
more
and
we
call
the
thing
to
sing
research
group,
so
you
know
it's
like
kind
of
like
something
coming,
my
nose
at
that
a
little
bit,
but
also
to
figure
out
well
what
we're
really
doing
is
orchestrating
things
with
logic,
so
the
other
thing
that's
missing
is
how
do
we
describe
logic
in
the
way
that
logic
connects
to
things?
D
D
D
C
Yeah,
the
annoying
part
was
that
well
synthetically.
We
had
to
do
it
and
mixing
up
protocol
and
application
logic
layers
because
we
had
to
embed,
but
that
was
a
syntactic
issue
that
we
realized
that
a
lot
of
errors
are
like
the
servers
down.
So
that's
not
really
an
application
level
problem.
It's
like
an
infrastructure.
C
F
F
F
The
discussion
on
in
our
july
think
of
last
year,
with
all
the
implies
that,
if
you
remember,
we
probably
don't
at
that
point,
we
made
a
proposal
to
write
a
paper
that
would
sort
of
list
the
criteria
for
how
to
describe
standards
and
what
they
are
trying
to
do,
and
also
what
they
are
assuming
implicitly
or
explicitly
in
terms
of
environment
and
node
capability
and
what
happens
at
design
time
versus
what
happens
at
prime
time
and
what
nodes
can
describe
and
what
nodes
can
discover.
F
F
And
I'm
beginning
to
think
of
this,
and
we
said
we're
going
to
just
do
the
high
level
description,
not
the
details
of
the
standards
themselves,
in
other
words
not
to
redescribe
them,
because
it's
demanding
and
not
adjustment,
whether
it's
good
or
bad.
It's
basically
what
it
is
and,
as
I
was
thinking
about
it,
it
almost
may
be.
A
partially
good
analogy
is
to
think
of
the
food
labels
that
you
get
on
the
full
product.
F
So
there
is
a
standardized
ways
of
of
the
amount
and
how
many
calories
and
what
are
the
nutritional
ingredients
that
the
value
and
stuff
like
that.
But
at
least
it's
in
a
way
that
can
be
compared
as
opposed
to
the
way
before
that.
But
everybody
starts
with
a
different
quantity
and
then
comes
up
with
different
numbers
anyway,
so
we
presented
the
comment
of
the
common
criteria
at
that
time.
F
They
didn't
get
much
discussion
that
it
turned
out
at
the
time
that
michael
costa
and
michael
mccool
had
worked
on
comparing
various
standards
along
some
common
criteria
in
terms
of
attitudes
and
features.
You
know
what
they
do
and
how
they
do
it,
and
there
are
some
cables
in
existence
from
both
at
the
various
degrees
of
completion
that
were
offered
as
a
contribution
to
that
work
I
haven't.
F
I
haven't,
had
the
chance
to
synchronize
with
michael
k
on
at
mccool,
but
in
a
moment
it's
going
to
become
clear
that
you
know
we
can
defer
that
for
the
time
being,
so
it
turns
out,
as
I
was
thinking
about
it-
and
you
know
a
reason
why
I
didn't
quite
get
off
the
ground.
Is
that
the
goal
of
describing
you
know
providing
all
of
this
that
I
said
plus
describing
some
key
standards
along
those
lines.
F
You
know
as
an
existence,
proof
and
a
comparison
turned
out
to
be
fun
to
have,
because
there
are
many
intricate
details
and
things.
So,
basically,
I'm
proposing
a
revised
approach
to
write
a
sort
of
position
document
that
describes
the
criteria,
how
to
compare
them
all
the
other
things
we
discussed
and
then,
rather
than
describing
standards
with
which,
as
I
said,
turned
out
difficult,
but
it
might
interest
the
sdo's
if
they
want
to
describe
how
their
standard
physical
and
we
could
use
apprentices
for
those
if
and
when
we
have
them.
F
But
the
point
is,
instead
of
going
for
the
whole
thing
criteria
plus
descriptions,
let's
start
with
the
criteria
as
a
position
and
do
the
same
thing.
Basically,
I
I
this
type
really
volunteered
to
do
the
first
draft
that
we
can
discuss
and
thrash
and
elicit
feedback
of
this
workgroup
and
revised
the
paper
produced.
The
approved
document
approved
either
says
that
the
group
says
okay.
F
This
is
how
far
we're
going
to
get
with
it,
not
necessarily
published,
and
then
invite
standards
seos
if
they
want
to
contribute,
and
if
we
are
successful,
we
can
use
the
feedback
from
them.
Also
revise
the
criteria
paper
because
we
will
need
adjustment
and
without
without
those
descriptions
published
the
paper,
I
even
went
through
the
visual
type
process
of
doing
xml,
rfc
and
all
the
other
good
stuff,
so
it
can
be
basically
an
rfc-like
format,
even
the
draft
proposed
so
operationally.
F
Basically
I
I
said
I
I
have
finally
waited
the
time
to
do
the
outline
of
this,
and
then
we
can.
You
know
we
can
discuss
it.
It's
a
concrete
thing
rather
than
a
concept,
and
I
don't
know
whether
it's
in
front
of
just
logistics,
the
rfc
format,
is
probably
what
it's
going
to
end
up
being
it's
a
little
bit
harder
to
edit
than
the
word,
but
probably
is
better,
but
not
in
that
format.
Custom.
You
think
we
should
barely
see
the
format
from
the
beginning
or
leave
it
from
it.
B
F
Yeah
yeah,
I
I'm
aware
of
it.
I
haven't
looked
at
it.
I
was
using
the
xml
to
rfc
macros
in
the
beginning,
in
fact,
to
get
that
to
work
with
it
anyway.
So
we
find
we
can
do
that
offline
and
so,
as
I
said,
I'm
volunteer,
I
think
it's
gonna.
F
I
have
to
travel
plan
to
europe
late
april
between
now
and
then
I'll
definitely
have
a
draft
and-
and
let's
just
put
it
on
it
in
meetings,
if
we
have
it
and
if
there
are
any
volunteer
co-authors,
that
would
be
great
if
not
dropped
out
and
then
we'll
take
it
from
there.
F
F
Manageable
and
then
we
can
decide
whether
it's
worth
pursuing
anyway,
so
yeah.
I
think
it's
an
easier
task.
C
F
Yes,
those
who
are
interested
just
escape
to
my
attention
recently
there
is
a
white
paper
from
iic
characteristics
of
iiot
information
models,
try
to
do
the
similar
thing
and
describe
standards
along
those
lines,
and
when
you
read
it,
you
find
out
the
difficult
tasks
for
this
approach,
maybe
by
the
way.
D
Nicholas
is
has
just
authored
a
white
paper
that
discusses
that
and
is
nicholas
on,
maybe
not
okay.
I
guess
not
anyway.
C
E
Iic,
industrial
internet
consortium
on
information
models
and
comparison
and
meta
models
and
making
these
collaborating
industrial
environments,
and
we
have
a
description
of
of
both
ipso
and
one
dm
there
together.
Other
folks
wrote
about
opc,
ua
and
so
on.
So
that's
maybe
something
to
look
at
as
well
bring
into
this.
F
Yeah
I
have
and,
as
I
said,
I
would
suggesting
that
here
we
defer
the
discussion
of
specific
standards
they're
in
because,
as
you
see,
you
know,
as
you
know,
it's
difficult
to
get
dispersed.
So
I
I
I'll
start
with
the
criteria
first
of
this,
but
I
think.
A
Excellent.
Thank
you.
Thank
you.
Unfortunately,
we're
running
over
time.
Now
we
need
to
do
the
hackathon
coronation.
One
thing
what
we
could
do
I
mean
we
could
briefly
mention
about
this
in
an
upcoming
filming,
rg
summary
meeting
that
could
also
you
know,
get
you
have
extra
volunteers
at
least
for
reviewing
this.
I
think
this
very
interesting
piece
and
I'm
very
good
if
you
and
mccool
and
other
interested
can
get
a
you
know,
have
a
side
meeting
and
see
the
best
way
to
progress
this.
A
A
Perfect
well
with
that,
we'll
have
to
delete
the
future
wikimedia
planning
offline,
but
thanks
a
lot
for
everyone
joining
today.
This
was
a
very
good
discussions.
Looking
forward
to
discuss
with
you
all
more
soon
again-
and
I
think
many
of
you
will
be
joining
for
the
academic
coordination
call
started
already
one
minute
ago,
so
see
you
on
that
side.