►
From YouTube: IETF102-NETMOD-20180720-0930
Description
NETMOD meeting session at IETF102
2018/07/20 0930
https://datatracker.ietf.org/meeting/102/proceedings/
A
To
pad
you
can
find
the
easiest
way
to
get
there
is
through
the
tools
page.
So
if
you
go
tools
that
ITF
org,
slash,
WG,
/,
net
mod
and
click
on
minutes,
you'll
find
etherpad
for
102
and
we'd
really
appreciate
everyone
in
the
room,
helping
to
contribute
to
those
notes
in
real
time.
I
think
that
work
has
started
the
session.
Welcome
to
our
second
net
mod
session.
This
is
the
last
session
for
us.
Obviously,
if
the
IPF
there's
one
more
session
after
this,
but
who
won't
be
there.
Thank
you
for
sticking
with
us.
A
We
have
about
a
half
full
agenda
and
I
know.
There
was
one
additional
discussion
topic
that
we
were
expecting
in
AV
to
talk
about
and
I
think
he
stepped
out.
So
he'll
come
back
in
and
we'll
have
that
conversation
when
she's
ready,
so
I'm
Lou
burgers
can
Watson
George
a
green,
and
we
have
Michael
taking
notes
here
as
well
as
our
secretary,
given
that
this
is
the
last
day
of
the
of
the
meeting
I
think
everyone
should
be
familiar
with
the
note.
A
We
are
sending
and
recording
video.
We
do
have
people
remote,
it's
very
important
that
if
you
speak
of
the
mic
to
say
your
name
and
I
apologize
if
I
forget
to
do
that,
we
have
blue
sheets,
circulating
sign
knows,
and
we
already
talked
about
ether
pad
and
our
agenda
is
online
and
unmodified.
This
is
the
second
session
you're
here
we
don't
remind
you
that
it's
there.
A
A
C
C
C
D
B
E
B
C
B
B
F
What's
in
Cisco,
and
so
I
like
to
think
about
such
things,
okay,
trans
con
I-
think
it
makes
sense
to
maybe
clarify
when
the
schema
differs
between
two
dates
stores.
What
that
means.
So
I
don't
have
the
draft
already
says.
That
means
the
data
we
were
talking,
yeah
excluded
from
the
comparison
if
there
is
potentially
one
node
in
one
scheme
and
once
again
around
so
clarify.
F
Is
just
on
the
data
instances
and
use
comparing
whether
the
values
are
there
in
the
to
date
stores
just
not
given
to
me
that
if,
for
example,
what
racial
state
this
or
does
not
implement
some
of
those
is
daeviated
remove,
then
you
want
to
always
flag
was
being
diffs.
So
I
think
you
need
to
just
consider
that
sort
of
case
and
maybe
have
another
option
to
say
whether
or
not
that's
reported
or
not.
I
could
talk
to
you
offline
and
then
one
other
Connie
can
get
back
to
your
original
slide.
F
G
I
C
B
C
D
C
E
C
C
I
J
A
C
C
C
F
Bob
Wilson
Cisco,
so
I
prefer
there
is
not
not
in
this
document,
so
I'm
not
debating
whether
or
not
and
documenting
the
capabilities.
Useful
actually
do
things
useful
thing
to
do,
but
I
have
keep
this.
This
draft
more
abstract
in
terms
of
this
is
just
how
you
do
this
data
and
you
use
the
capabilities
as
an
example
effectively
in
this
draft.
F
That's
fine
I
want
to
have
a
separate
draft,
the
talks
about
device
capabilities-
and
it's
probably
more
stuff
to
say
about
them
and
says
that
the
way
to
do
is
use
instance,
data
documents
and
many
of
you
in
your
previous
line.
You
said
about
comparing
like
old
data
and
the
current
one
again,
that's
that's
used
to
be
thing
to
do
with
capabilities.
F
But
then
I'll
describe
this
you're
saying
this
is
just
providing
a
snapshot
of
data
at
one
point
in
time,
but
I
mean
it's
of
course
a
document
is
this
written
to
a
file,
can't
obviously
update,
and
it's
the
context
of
how
it's
used
defines,
which
one
is
the
right
one
to
use.
I,
don't
think
this
draft
should
say
that
it's
inter
strange
thing
to
say
this
is
just
about
how
you
represent
long
instance.
Data.
C
K
B
B
A
M
Joke
like
Cisco
so
as
part
of
the
work
in
the
main
catalog
for
the
in
catalog,
every
Cisco
has
been
trying
to
do
even
before
this
existed,
publish
some
of
the
device
capability.
In
fact,
probably
the
hello
message
is
in
github
I'm,
not
saying
that
this
that's
how
they
should
that's,
how
the
how
should
be
implemented.
I'm,
just
saying
that
when
you
presented
this
I
thought
this
is
the
format
we
could
use.
N
C
N
Okay,
so
the
the
fact
that
the
factory
store
exists
just
connecting
dots,
so
the
the
instance
data
is
what
you
would
expect
from
you.
You're
manufacturing
element
right,
yes,
okay
and
it's
it's
provided
outside
the
context
of
a
deployed
element
right,
yes
into
some
location
that
you
can
obtain
that
contains
a
config
data
right
website.
So.
N
J
F
Just
trying
to
connect
all
of
us
together
so
robots
in
Cisco,
I'm,
not
sure
really.
The
answer
I
overthought
was
in
the
factory
store,
is
just
the
default
configuration
where
the
device
comes
up.
So
if
you
just
power
on
for
the
very
first
time,
that's
the
configuration
you
get
that
gets
written
to
start
off
or
running
where,
as
surely
the
capabilities
data
is
just
about
what
functionality
the
device
can
do,
protocol
enabler
features
you
can
have.
E
N
C
N
G
F
Well,
but
since
Cisco,
so
I
just
have
one
on
the
common
that
the
capabilities,
inflation
itself
I,
don't
think
you
get
interoperability.
If
you
just
say
you
need
to
publish
this
data
needs
to
find
one
level
of
what
that
looks
like
that.
You,
otherwise
just
saying
everyone
should
do
this,
I'm,
not
sure
how
that
helps
the
world
alone.
I.
C
F
So
I
think
it
needs
definition
of
these
definition
of
needs
to
be
yeah
model
defines
what
this
capabilities
data
looks
like.
What
does
it
contain?
How
do
vendors
extend
it,
because
otherwise,
it's
just
I
get
some
random
beta.
You
should
provide
some
random
beta
that,
in
some
format,
I'm
not
sure
that
that
helps
friend.
F
C
A
A
You
have
to
find
what
that
instance.
Data
is
so,
if
you
know
I
think
if
you
want
to
try
to
keep
this
you're
gonna
have
to
define
what
the
format
is
that
you
want
to
publish
you,
don't
necessarily
have
to
format
and
publish
I
include
the
transport
in
this
document.
You
can
have
that
elsewhere,
but
you
least
have
to
say
what
the
information
is,
that
you're
publishing.
K
From
Jim
for
networks
is
following
up
on
ten
carries
question
I,
think,
there's
quite
a
difference
between
this
and
the
factory
data
store
effort,
the
factory
data
search
and
capture
the
configuration
that
we're
going
to
start
up.
We
said
well
idea:
this
is
primarily
capturing
well
with
being
library
and
young
library
is
purely
an
up.
My
data
store,
so
it's
not
actually
even
wouldn't
be
found
inside
the
factory.
C
But
it's
more,
for
example,
we
already
use
this
for
defining
a
predefined
access
control
rose,
Angeles,
which
is
quite
a
big
work
to
do,
and
it's
good
to
help
and
that's
configuration.
You
could
also
use
this
for
documenting
that
that's
factory
default
comes
up
with
one
internet
interface
configure
to
whatever,
so
it's
really
free
of
use,
but
yes,
you're
right
that
much
of
this
will
be
about
state
data,
read-only
data.
So
it's
more
a
superset
and
yes.
P
P
Okay,
so
at
the
moment
there's
no
overlap,
but
I
see
the
trajectories
overlap
in
some
time
in
the
future.
So
mud
is
talking
about
advertising
what
what
a
device
is
and
what
its
network
requirements
are,
whether
a
sentence
that
says
and
in
the
future
we
will
add
a
description
of
more
things
that
the
device
or
the
server.
What
other
can
do
and
I
see
this
as
possibly
intersecting
somewhere
two
years
in
the
future
or
three
years.
I
C
Q
C
C
We
should
reuse
that,
but
I
don't
see
using
radiation
Bates
as
a
viable
solution,
because
most
of
the
times
instance
data
doesn't
change.
When
you
make
a
small
addition
to
your
module,
that's
potentially
what
I
see
is
maybe
a
list
of
modules
and
revisions
that
that
is
usable
with
this
instance.
Data
set
okay,
know
by
your
favorite
question.
I
Yang
data
extension,
it
just
defined
some
schema,
so
if
an
offer
of
a
server
for
whatever
reason
decides
to
implement
it,
I
don't
see
anything
wrong
so
right.
So
it's
and
the
practical
consequences
of
doing
so,
of
course,
are
true.
First,
you
cannot
use
tools
for
processing
this
yeah
and
what
yoga
do
for
this
extension
and
I
suspect.
This
is
why
you
didn't
include
Eska
tree
of
this
module
in
your
document,
because
you
can,
you
cannot
print
it
with
pay
angle.
I
C
I
H
A
M
A
M
H
C
C
C
F
Robots
in
Cisco
and
so
I
think,
if
the
dates
approach
was
included,
it'd
be
an
optional
thing,
so
you
could
include.
It
was
useful,
even
have
to
populate
where
it
is
I
think
it
was
to
be
defined.
So
I
remember
guys
when
it's
mandatory,
it
might
be
places
where
as
useful
outside
I'm,
not
sure
the
estate
below,
in
terms
of
it
being
confiscated
once
operatio
and
convict
true
bronze
mine
calculate
is
that
is
that
can
actually
work.
Isn't
this
just
data
coming
out
of
a
datastore,
so
you
have
a
schema
node,
that's
being
this
conflict?
F
C
F
Might
use
it
differently?
Okay,
so
give
a
concrete
example
say
you
had
a
list
of
interfaces.
That's
the
data
you're
returning
in
nmda.
Those
nodes
are
config.
True,
so
you're
saying
that
by
definition,
those
list
of
interfaces
have
to
be
configuration
because
the
scheme
Association
is
title
true,
whereas
they
might
be
in
spaces
that
just
instantiate
in
the
system
haven't
been
configured
so
I
I.
Think
you
need
some
more
indication
as
to
what
the
data
is.
C
F
Just
providing
context
to
allow
some
who's
interpreting
the
data
to
understand
what
it
means
I
mean.
Does
this
mean
configuring
you're,
saying
automatically?
That
will
mean
configuration
because
it's
because
those
nodes
are
defined
in
the
scheme
was
conflict
true,
but
but
a
half
the
data
in
operations
could
be
defined
as
conflict.
True
and
you're
now
saying.
Actually
all
that
data
is
not
state
data,
that's
all
configuration
data,
because
it's
convict
reunites
doesn't
make
sense
to
me.
F
C
A
Times
it
seems
like
we're,
mixing
some
concepts
which
are
really
the
data
representation
focused
on
build
a
representation
and
then
other
concepts,
items
that
are
focused
on
how
we
are
moving,
that
representation
or
restoring
that
representation,
and,
just
like
we
have
the
separation
between
that
mod
and
Netcom
for
people
to
keep
that
separation
in
this
work.
So.
C
C
E
C
C
The
next
one
was
do
we
need
to
create
tags
like
we
have.
Code
begins
and
code
ends
today
for
hearing
modules
and
for
other
things
should
we
if
we
would
use
such
X
examples?
For
instance,
data
begins
and
instance,
data
ends
in
a
draft
or
an
RFC
I
would
save
no,
because
this
would
be
typically
used
for
examples
in
drafts
and
even
for
the
end
modules.
We
don't
use
the
code
begin
and
code
and
tags
and
we
are
pulling
the
examples.
We
might
not
need
the
record
about
the
instance.
C
J
C
J
I
I
Probably
these
files
will
often
be
stored
in
some
kind
of
version
management
system,
I
get
and
for
such
systems.
This
file
name
scheme
is
next
to
unusable
because
a
relevation
you
have
to
do
something.
So
we
can
just
committed
a
cities,
and,
given
that
this
data
are
stored
as
metadata
inside
the
file,
it
shouldn't
be
needed
to
use
the
file
name
for
encoding
and
information.
So
I
would
suggest
maybe
two
if
this
prescription
completely
just
the
ferric
for
nine,
can
be
and
I
think
it's
not
so
important.
C
Well,
that
and
95%
of
the
cases
we
don't
use
their
graduation
date
in
language
modules
either.
In
some
cases
it
is
good
and
if,
in
some
case
it
is
good
that
the
tools
will
allow
it
I
thought
following
the
yank
breakfast.
Maybe
we
were
and
adding
comments
about
version
control
would
be
good,
but
is
that
ok,
materials
right?
Ok,
then
I
read
know
that
usually
avoid
because
of
Russian
control
problems.
A
A
A
A
A
B
B
Use
is
consolidated,
job,
that's
a
basis
to
ensure
that
it's
cutting
the
strands
apart
and
the
change
we
made
actually
compared
with
the
previous
vision.
Actually,
we
define
the
14
structures
such
as
rather
structure
only
structure,
and
why
not
with
them
include
the
photo
structure.
Yeah
there'll
be
a
shoe
everybody's
container
and
we
also
have
sculptor
actually
put
on
appendix
and
we
have
the
additional
cancer
and
remove
the
top
pastor,
based
on
the
Menenius
that
it's
cause
she
and
we
somehow
it's
constant
on
the
lister.
We
won't.
You
will
pop
back
to
of
the
issue.
B
R
B
We
use
the
toolspace
upset
so
just
like
Aziz
example
before
the
forty
might
have
you
know,
so,
there's
no
mystery
here
now,
after
floating,
they,
actually
we
can
reduce
speeds.
Are
the
Nona
in
two
to
nine
independent
second,
in
line
with
two
space
from
the
left,
so
the
benefit
to
take
this
powder
is
this
because
we
use
Bacchus
last
year
to
speed
on
mine,
so
this
kind
of
backslash
placement
can
be
the
fixed
or
verbal,
and
so
this
actually
has
been
discussed
on
the
list
and
we
sing
the.
B
B
B
B
F
G
F
P
Actually
got
up
when
you
said
the
word
examples,
so
I'm
co-author
of
this
and
I've
been
a
little
bit
of
sleep
at
the
Lord
for
the
last
few
words,
but
in
my
opening
my
editor
I've
got
a
pile
and
examples
that
I
want
to
drop
into
the
draft
I'm,
not
sure
that
those
examples
help
resolve
that
question.
That
Chuy's
are
asking
about
the
different
options.
P
But
my
concern
is
not
just
Auto
full
fold
in
which
I
think
was
where
Kent
was
coming
from.
As
I
write
my
drafts
in
XML
and
handcrafted
XML
work,
my
father's
father
did
and
when
I
do
that
I
sort
of
automatically
insert
the
indentation
spacing
that
I
like
to
see,
although
I
want
to
create
in
the
output
text
file
and
therefore
the
only
floater
has
to
suppose
that
and
I
think
that,
but
whichever
option
we
choose
for
the
auto,
for
whatever
we've
got
a
support,
the
manual
profiting.
F
So
because,
in
terms
of
these
examples,
you
hear
backs
proposal
three,
that's
the
one
I
put
forward.
F
We
also
have
option
here
of
right,
indent
in
that
second
line
wherever
you
like,
and
you
know
you
can
strip
all
that
space
in
between,
whereas
if
you
have
one
of
the
cases
where
and
you're
not
using
fixed
amount,
what
happens
when
you
to
a
splitter
or
whitespace
point,
then
then?
How
do
you
do
that?
So
I
think
one
that's
floating
doesn't
work
unless
you're
some
way
of
denoting
where
it
finishes.
I
Draft
in
the
transport
on
be
our
design
team,
where
we
are
developing
a
lot
of
JSON
code,
and
we
found
it's
very
useful
and
the
one
comment
I
got
from
the
members
of
one
team
is
that
they
will
have
to
prefer
to
see
the
folding
being
done
at
the
end
of
the
world
rather
than
in
the
middle
of
the
word
for
readability.
It
will
be
much
more
readable
if
you
fold
again
when
you
have
a
space
rather
than
just
in
the
middle
of
the
line.
If
that
is
possible,
will
be
a
good
announcement
and.
K
O
K
See
because,
ideally
the
would
be
for
the
intimate
tool
for
thinking
being
that
only
text-based
presentation
format
needs
to
be
in
photos
if
the
presentation
format
is
HTML
or
something
else
and
world
would
not
want
to
for
it
to
be
folded
and
and
similarly
for
extraction
purposes.
If
it's,
for
instance,
the
XML
format,
you
don't
need
to
have
any
tags
for
do
an
extraction
they've
already
already
inside
tags.
So
perhaps
it's
a
dimensional
speak.
K
A
With
with
definite
contributor,
but
another
hat,
not
sure
which
had
I
think
we
should
decide
what
we
want
to
be
done
and
then
asked
willing
to
do
it,
I,
don't
think
we
should
be
constrained
by
what
the
tooling
is
today.
The
tools
are
generally
changeable.
Let's
figure
out
what
the
right
answer
is
and
then
also
figure
out
where
we
documented,
whether
it
be
a
standalone
document,
or
this
document
or
somewhere
else,
but
I
would
strongly
encourage
not
talking
to
the
tools
team
until
we
know
what
we
want
to
do.
P
A
P
A
J
J
K
J
N
Because
I
thought
we
talked
about
this
last
time
we
get
I,
think
it's
I
go
with
Andy
and
in
the
sense
we
know
we
have
the
problem
but
I
think
between
here
and
there
was
a
finite
state.
There
was
a
couple
of
ways
of
doing
this
that
that
we
were
still
trying
to
figure
out,
and
then
there
was
some
direction
from
101.
That
said,
hey
you
guys
get
together
and
see
if
we
can
come
in
with
with
a
decent
solution
or
not
I,
don't
know.
If
that
happened,
I
yeah
we
actually
had
to
get.
N
A
J
A
Jabar
yeah,
you
go
ahead,
okay,
thanks
so
from
from
jabber
you're
gonna
ask
the
question
is:
does
the
working
group
have
the
resources
to
work
on
things
other
things
we
take
on
say
that
what's
a
I
think
a
chair
question
as
chair
my
views
or
contribution
driven.
So
if
people
bring
things
to
the
working
group
and
the
working
group
is
interested
in
expending
energy
there,
that's
that's
what
we
take
on
and,
of
course,
I've
cooked
yeah.
L
Joey,
not
speaking
as
a
chair,
I,
think
election
would
and
I
think
they
were
a
little
weird
for
us
to
progress
this
one
and
the
one
that
comes
behind
it,
like
they
sort
of
seem
related
to
each
other
sufficiently
that
like
mmm-hmm,
it
looks
like
a
working
group
where
the
working
group
participants
are
exploring
the
problem
and
solution
space
and,
like
there's,
probably
a
synthesis
of
all
the
work.
That's
being
done
around
this.
L
That
comes
out
of
that
great
I'm,
not
sure
that
that's
a
particular
document,
the
idea
of
working
in
this
space
seems
appropriate,
but
I'm
not
sure
either
this
document
or
the
subsequent
one
are
necessarily
like
the
place
that
we
should
be
when
we
get
to
adopting
something
and
I
say
that
as
a
contributor
and
or
as
a
idiot
and
the
working
group
and
not
as
someone
with
a
great
deal
of
experience
or
with
respect
to
channeling
the
chairs
opinion.
A
A
K
Also
regarding
question
I
understand
that
what
contribution
driven,
I'm
suppose
there's
interest
the
contribution,
that's
Anna
Freud,
but
also
dozen
sometimes
there's
no
notion
of
prioritization.
You
know
holistically
like
many
different
things
independently,
but
holistically,
then
there's
something
from
others.
I
mean
it's
not
really
unique
to
this
working
group,
but
I
think
every
working
group,
so
perhaps
others
with
more
experience
or
80s.
Could
how
do
other
work
in
groups
address
this
problem.
G
A
Here
I
mean
I've,
been
in
other
working
groups
where,
basically
it's
you
know
as
more
contributions
come
in
and
you
actually
can
do
more
work,
because
there's
now
more
people
able
to
work
in
parallel,
so
manure,
--xx
out,
okay
and
that
and
the
workload
can
dramatically
increase
that
doesn't
always
work.
So
we
have
to
figure
out
what's
right
for
this
group,
but
you
know
generally,
our
processes
is
being
driven
by
contributions,
not
chair,
not
we're
sort
of
well
thought
them
up
and
top
down.
G
It
was
known
as
that
you
asked
so
these
two
documents
seem
to
be
focusing
more
on
the
actual
problem
space
of
usability
than
on
the
paradigms
themselves
and
for
someone
original,
reversing
opps
awgn
came
here
because
the
feeling
that
was
that
it
is
generic
enough
to
be
used
as
a
generic
mechanism
for
modeling
the
state
machines.
I,
don't
have
a
strong
feeling
whether
this
should
be
done
here
or
in
some
other
working
group.
E
R
R
Elements
so
for
frame
we
denta
fide
some
use
cases.
One
is
brain
struct
data
play
advices
of
an
optical
network,
for
example,
does
optical
transponders
two
about
configuration
settings
in
case
of
physical
layer
changes
so,
for
example,
in
case
of
physical
layer
definitions,
the
objective
to
install
this
phoenix
state
machine
in
the
agent
of
the
transponder?
To
that
something
happens,
the
transponder
the
agent
already
knows
what
to
do
and
it
can
apply
the
reconfiguration
settings
so
based
on
the
physical
conditions,
though
the
proper
washing
formats,
the
coding
and
so
on.
Another
useful
is.
R
Telemetry
and
another
one
with
that,
we
identified
is
related
to
to
the
monitoring
of
of
packet
loss.
So,
regarding
a
quest
question
related
to
the
previous
to
the
previous
talk
about
the
implementation
we
implement
ich,
finished
and
tested
in
lab
and
also
in
a
field
trial
at
Telecom
Italia,
for
what
concerns
the
first
use
case.
R
So
there
a
configuration
of
the
of
the
of
the
transponders
in
in
an
optical
network,
so
the
v
trial
was
successful
and
the
transponder
was
able
to
automatically
reconfigure
without
acquiring
or
asking
something
through
the
to
the
Sdn
controller,
because
it
was
a
prayer
instantly.
So
we
next
next
slide
please
so
this
slide
describes.
J
R
R
Next
slide,
please
the
second
use
case
if
work
at
telemetry
here,
the
the
objective
is
to
reduce
telemetry
bandwidth
on
demand
data
collection.
So
in
this
case
we
we
can
see
network
probes,
as
as
a
phoenix
state
can
be
described
that
word
probes
as
a
phoenix
state
machine
so
based
on
the
state,
we
we
send
data
stream,
Telemachus
team
to
to
the
network
telemetry
applications,
so
we
can
optimize
the
bandwidth
drop
of
the
telemetry
state.
The
third
use
case
next
slide.
Please
is
related
to
I.
R
E
E
R
The
keys
of
the
transponder,
each
state
can
correspond
to
two
specific
settings
of
setting
of
the
transmission
parameter
of
a
transponder
and
to
each
state.
A
list
of
transmission
is
associated
to
each
state
and
we
can
go
from
a
state
to
another
state
when
something
that
change
so,
for
example,
in
the
case
of
the
optical
network,
one
deteriorate
increase.
R
Then
each
transition
is
associated
to
a
list
of
actions
that,
for
the
case
of
the
transponder,
could
could
be
the
change
of
the
modulation
format
and
the
change
of
the
code
rate
and
and
so
on.
So
this
list
of
action.
We
have
not
execute
that
when
whether
that
actually
implement
implements
change
the
actual
and
then
we
can
go
to.
R
R
M
A
R
Well,
I
have
two
two
impressions.
My
face
is
my
first
impression
is
that
this
draft
is
more,
is
more
generic
than
than
the
than
the
other
one,
and
then
I
have
another
impression,
but
I'm
not
sure
about
this.
I
have
to
better
understand
the
other
draft.
Maybe
we
can
have
an
offline
discussion
with
the
other
authors,
but
my
impression
is
that
their
draft
is
more
associated
to
to
central
control.
While
here
the
objective
is
to
give
a
sort
of
intelligence
to
to
the
a
to
the
agent.
E
E
A
L
H
H
We
reviews
our
health
system
young
module
page
now.
It
again
covers
some
basic
function.
Those
are
as
users
as
notation,
but
it's
a
lick
of
feature
of
some
punctures
to
support
as
a
user
authorization,
accounting
and
also
the
current
system.
Young
model,
not
a
supported
caucus
process
server.
So
we
propose
this
young
motor
to
address
these
issues
and
also
I
know
a
lot
of
window.
Has
employment
attacks
passed
in
their
device
so
and
they
also,
we
have
say
average
under
the
module,
so
I
think
it
may
be.
An
issue
should
be
discuss.
H
B
H
H
E
H
L
H
M
So
I'm
gonna,
let
him
speak
jokes
are
playing
I'd
say
WG
co-chair
had
so
yes,
as
was
brought
up,
we
mentioned
it
in,
say:
WG.
We
didn't
have
an
agenda
slack
for
it.
I
agree
that
drove
everything
you
said.
Essentially,
we
are
trying
to
document
how
tax
is
implemented
today
in
the
industry
and
get
vendors
who've,
implemented
clients
and
servers
to
feedback
into
that.
M
G
Donors
to
give
a
perspective
of
the
tax
work,
the
current
work
is
to
document
what
is
there
not
to
work
on
the
additional
features
and
other
things?
Well,
this
needs
to
be
finalized.
At
tactics
is
widely
deployed
and
that's
how
commodity
service.
So
from
a
perspective
of
that,
it
is
used.
Yes,
certainly
a
model,
for
that
seems
to
be
needed.
G
One
concern
that
I
will
have
is
that
it's
not
only
the
tactics,
it's
also
the
radius,
which
is
frequently
used
for
similar
things,
and
if
we
look
from
a
system
perspective,
those
two
protocols
can
be
selected
by
the
choice
of
the
deployment
of
a
basic
would
buy
it
by
the
user.
So
maybe
it
would
be
a
good
idea
to
try
to
work
a
little
bit
on
the
overall
external
authentication
authorization
system
framework
and
then
have
tactics
and
radius
put
into
that.
G
H
Why
we
applauds
they
start
opening
to
the
molding,
because
we
think
it
see
arts
and
some
function
awesome
attribute
should
be
in
she.
Should
she,
because
I
was
I'd,
have
sister
motive
way,
since
it
can
argument
I'm
extensions,
a
sister
might
have
sister
module
and
to
cover
some
attribute
or
some
feature
not
currently
not
covers
just
why
we
want
you
avoid
white.
Why
way
synthesis
drafted
to
current
what
improvements
so
also
I.
G
A
A
With
the
comment
baby's
coming,
okay,
thank
you
also.
I,
think
that
they
you're
talking
about
is
that
the
common
framework
makes
a
lot
of
sense.
Would
you
think
that
the
common
framework
still
belongs
over
there,
but
I'm
happy
with
either
answer
actually
I,
probably
think
that
the
common
framework
might
be
better
tied
to
the
actual
protocols?
So,
what's
your
recommendation.
A
G
Guys
can
have
questions,
because
if
we
are
looking
at
the
practical
system,
how
that
user's
authentication
sources,
it's
not
only
tactics
and
radios
the
mind,
the
local
store.
There
might
be
something
else,
and
if
we
look
this
way,
it
quite
quickly
gets
into
the
napkin
keystore
topics.
It
gets
into
other
key
store
topics.
I,
don't
have
an
answer
for
that
right
now,
but
if
we
look
from
the
user
perspective,
suddenly,
flooding
different
mechanisms
for
using
similar
functionality
is
not
the
right
thing.
G
L
Joey
way
as
a
chair
at
this
point
like
if
there
were
extensions
to
tax
that
were
to
come
in
the
other
model,
I
would
think
that
would
possibly
be
inappropriate
and
dangerous
at
this
point,
since
we're
still
working
on
documenting
tac-x
itself
on
but
like
those
extensions
would
definitely
be.
This
is
the
sort
of
thing
that
ops
AWG
would
be
looking
at
in
light
of
the
fact
that
they
actually
sort
of
owned
the
progressing
of
of
tackett's
itself,
so
I
wouldn't
want
to
I,
wouldn't.