►
From YouTube: IETF113-ROLL-20220323-1200
Description
ROLL meeting session at IETF113
2022/03/23 1200
https://datatracker.ietf.org/meeting/113/proceedings/
A
The
concert
house,
the
opera-
I
think
I
did
walk
past
it
then.
A
A
I
think
yes.
C
Yes,
I
know:
okay,
we're
gonna
start,
I
think
so,
let's
go.
Thank
you
very
much
for
joining
us.
This
is
the
itf
troll
meeting.
C
So
we
are
going
to
delegate
the
handling
of
the
sites
to
you
when
you
need
to
present
make
sure
of
the
resources
they
are
outlined.
B
I
do
this
part,
this
is
an
itf
meeting
and,
as
such,
you
need
to
be
aware
of
the
noteworth,
notably
of
the
ipr
policies.
If
you
know
of
any
ip
pertaining
to
any
of
the
topics
being
discussed,
videos
or
somebody
else's,
you
know
need
to
either
mention
it
or
not.
Talk
about
the
topic,
and
also
I
want
to
mention
the
harassment
topic
so
with
bcb25.
B
C
D
E
B
C
C
Okay,
thank
you.
We
proceed,
so
this
is
our
trap
status,
so
we
got
thank
you
alvaro
for
the
update
in
the
nsa
extension.
So
now
the
others
I
need
to
replay
and
address
typing
issues
then
for
our
developer.
It's
this
is
our.
This
document
has
open
issues
and
we're
going
to
discuss
next
on
this,
and
our
projections
are
going
to
be
presented
today
and
it's
a
kind
of
all
the
issues,
open
issues
or
addresses.
C
C
C
Okay,
so
for
milestones
well,
we
have
we
proposed
changes.
I
don't
know
dominic.
We
want
to
take
lead
on
this.
Okay.
B
B
So
here
are
the
milestones
that
were
defined
so
far
and
I'm
we're
proposing
new
dates.
So
we've
put
years
but
not
months,
so
can
we
have
an
estimate
for
those
and
so
that
we
can
refresh
the
milestones,
so
let's
go
through
them
their
projection.
D
F
Hi
alberto
with
a
rolling
d:
don't
progress
the
work
based
on
what
the
working
group
can
do.
My
my
queue?
Yes,
is
very
deep
right
now
and
I'm
running
behind,
but
don't
look
at
that.
You
know
just
you
know,
progress
to
work
as
the
worker
needs
to
progress
it
and
I'm
working
on
finding
a
way
to
drain
the
queue
a
little
bit
faster.
F
There
were
some
documents,
especially
last
year
that
were
bigger
documents,
not
from
this
working
group,
but
that
I
had
to
you
know,
process
and
they
you
know,
custom
blocking
there.
So
you
know
process,
as
you
normally
would,
and
I'll
try
and
take
care
of
this
faster.
E
My
take
is
we
had
more
than
one
very
deep
review
recently
and
you
will
see
in
my
updates,
but
yeah
I
think
we
are
we
are.
We
have
reached
the
quality
of
workgroup
call,
so
I
will
be
happy
when
you
start
it
and
and
my
status
was
remo.
Saris
was
supposed
to
give
one
last
review
before
walker
crystal,
but
since
we
got
one
for
from
lee
one
from
torres,
I
think
remoussary's
review
can
can
be
one
of
those
for
lasko,
as
opposed
to
before.
Let's
go.
E
B
E
Yeah
definitely
even
earlier
I
mean,
if
you
start
the
workplace
call.
Now
I
mean
as
soon
as
you
know,
the
current
last
goals
are
completed.
You
may
start
this
one
and
then
it's
a
matter
of
amount
of
max.
C
There
are
some
open
issues,
maybe
june
or
september.
What
do
you
think,
michael.
A
I'm
just
looking
and
I'm
thinking
we
we
should
fix
the
title.
Maybe
the
title
is
wrong,
but
yeah.
No,
I
I
actually
think
we
really
could
push
the
button
before
are
at
around
the
next
ietf.
I
think
so
I
think
I
mean
it's.
It
got
to
maturity
point
and
then
we
said
oops,
so
I
think
we
just
need
to
resolve
the
discussion
as
to
what
we
need
to
do
and
we'll
do
that.
C
A
I
I
think,
that's
maybe
reasonable,
but
that's
good.
It's
going
to
take
us.
I
think
we're
going
to
need
some
some
some
interim
meetings
to
to
get
to
that,
because
I
think
there's
still
a
lot
of
of
open
questions
that
there
and
and
it
yeah
it
would
be
great
to
have
some
implementations
to
do
something
like
this.
Our
team.
C
C
C
Okay
and
then
yes,
okay
for
young
model
friend
mpl,
do
you
want
to
work
on
this
topic?
I
mean
this
is.
C
Okay,
yes,
then,.
C
And
no,
it
was
by
a
long
time
ago.
It
was
driven
by
peter
right.
C
B
C
F
Yeah
hi
I'll
go
right
on
about
anything.
Yes,
so
the
the
part
of
the
required
process
is
the
charter
right.
We
need
to
change
the
charter
to
change
the
milestones.
F
You
know
you
change
them,
you
add
the
lead,
whatever
I'm
gonna
get
an
email
to
to
approve,
but
in
general
you
know
this
is
a
big
part
of
this
is
your
decision
now.
As
you
know,
I
said
the
asg
is
paying
specialization
to
manageability
and
the
normal
question
comes
up
with
every
draft,
so
where's
the
yang
model.
For
this.
How
are
we
going
to
manage
this
network?
F
The
one
thing
that
that
you
know
we
probably
or
you
need
to
discuss
in
the
working
group.
Is
you
know?
How
do
these
networks
get
managed?
You
know
is
yang
the
appropriate
way
to
do
it.
Are
there
other
tools?
You
know?
I
don't
know,
because
if
we
need
work
to
be
done,
you
know
we
need
to
work
work
to
be
done,
because
otherwise
the
solutions
wouldn't
be
complete.
F
Ideally,
we
would
find
people
in
the
working
group
that
are
interested,
so
maybe
that's
the
type
of
conversation
that
we
need
to
have,
because
not
all
say,
iot
related
deployments
are
going
to
have
the
same
type
of
manageability
that
other
networks
or
other
types
of
numbers
are
going
to
have.
So
if
yang
makes
sense
sure
at
some
point
we
need
to
do
it.
If
it
doesn't
make
sense,
then
you
know
maybe
not
so
we
can
then
negotiate
that
when
we
recharge
it
at
some
point
so.
A
We
don't
we
don't
have
a
a
work
item
right
now
as
a
yang
model
for
ripple.
Okay.
This
is
a
yang
model
for
mipple,
correct
right
and
I
would
say
that
that
there's
close
to
zero
deployment
of
nipple
today,
some
of
the
lighting
and
building
people
were
very
enthusiastic
about
it,
and
I
don't
know
if
they're
still
enthusiastic
about
it.
Maybe
that
that
may.
C
A
E
B
E
Okay,
so
I
was
saying
people
is
being
used
by
myself,
but
by
side
is
not
using
young,
and
until
we
have
an
actual
deployment
that
uses
ripple
and
my
our
people
had
decided
to
work
with
yank.
E
C
E
C
Okay,
we
will
take
to
confirmation
in
the
main
list
of
the
removal
of
this
work.
E
C
Okay,
thank
you
about
the
next
topic,
self-taught
multicast
for
ripple.
E
E
But
if
you
look
inside
the
review,
you
will
see
why
he
abstained
and
he's
basically
telling
us
that
the
work
group
has
not
done
before
work.
The
quality
of
the
document
was
not
good
enough
when
it
twitched
isg.
That's
what
ben
is
telling
us
I'm
paraphrasing,
but
please
look
at
his
words
and
I
was
asked
to
do
a
review
on
version
13.
E
I
did
I
published
it.
I
used
gentle
words,
but
no
sorry,
we
are
not
talking.
C
About
how
they
be
labeled,
it's
sort
of
they,
oh
I'm
sorry
just
was
not
against.
Oh
sorry,
I
chose
a
sauce.
E
Sorry,
oh,
no!
No!
No
progress
on
that.
We
just
discussed
it
with
gaston
this
morning
because
we
because
of
the
other
draft,
the
the
one
on
the
story,
a
non-starring
mode
multicast
which
which
I
will
be
discussing
so
he
said,
did
somebody
work
on
how
that
interacts
with
c-cast.
E
The
only
thing
I
know
is
probably
a
look
at
it
again
with
casper
to
see
how
it
integrates
with
the
multicast
registration
work
and
the
mach
5,
but.
B
C
Okay,
then,
okay,
we
are
going
to
be
fast
here.
There
are
open
the
issues
to
be
discussed
today
about
problem
and
enrollment
priority.
C
But
yes,
now
the
work.
The
this
document
have
changed
the
mode
of
operation
to
four.
G
C
C
E
So
we
cannot
push
the
button
for
publication
request
at
this
time,
not
at
all
and
whether
there
are
enough
hands
to
work
on
it.
I
I
I
could
do
one
more
review
in
the
future,
no
worries,
but
I
will
not
contribute,
and
so
we
need
active
contributor,
really
implement
and
look
at
my
commands.
It's
not
even
clear.
It
works.
C
Okay,
thank
you
very
much
pascal.
So
yes,
indeed
this
document,
they
need
to
be
addressed
all
the
open
issues
before
further
further
publishing
process.
E
E
F
Well
I'll
go
over
a
thunder
rolling
id
so.
F
So
this
document
already
went
through
working
with
last
call
and
iatf
last
call
and
asg
evaluation,
and
I
read
it
and
everything
else.
And
yes,
you
know
there
are
a
number
of
topics
that
ben
brought
up.
He
ended
up.
Staying
abstain,
usually
means
there's
so
much
work.
That
needs
to
be
done,
but
I'm
not.
I
don't
have
time,
I'm
not
willing
to
engage
in.
That
ben,
of
course,
is
stepping
down
today
as
security
ad.
F
So
you
know
he
doesn't
have
any
more
cycles
to
to
work
on
this
and
and
pascal's
recent
review
really
concerned
me
because
again
you
know
this
document
left
the
working
group
a
year
ago,
or
so
we
are
at
a
point
where
I
could
actually
approve
the
document,
there's
enough
positions
to
to
move
it
forward,
but
there
are
enough
holes
and
things
that
need
to
be
fixed,
that
that's
definitely
not
the
right
thing.
F
The
right
thing
to
me
seems
to
be
to
return
the
document
to
the
working
group,
which
is
where
I
think
that
the
question
from
ines
comes
of
of
was
there
interest
in
the
working
group
to
work
on
this
topic,
because
if
there
isn't
then
of
course
we're
almost
sentencing
the
document
to
death
right
because
we're
returning
it.
If
the
worker
is
not
going
to
work
on
it,
then
it's
probably
not
going
to
go
anywhere.
I
I
hate
to
see
work
that
people
have
done.
F
So
what
I
think
I
need
to
do
is
return
the
document
to
the
working
group
and
if
or
as
you
find
time
to
an
interest
to
work
on
it,
then
you
you
can
right
and-
and
if
it
ever
comes
back
to
me
and
the
aesg,
then
we'll
do
the
processing
at
that
point.
But
yeah
there's
enough
issues
outstanding
and
open
right
now
and
enough
concerns
that
that
I
I
can't
push
the
document
forward.
C
E
Okay,
okay,
so
this
is
about
the
the
dow
projection.
Just
for
those
who
are
new
to
the
topic,
the
the
main
people
standard,
rfc,
67,
65
550,
I'm
sorry,
is
a
classical
quote-unquote
distance,
vector
with
a
protocol
with
some
specific
adaptation
for
iot.
In
particular,
it's
an
is
appropriate,
meaning
that
there's
a
sense
of
direction
in
separation,
but
it's
it's
a
distance
vector
family.
Now
there
are
extreme
use
cases
where
it's
probably
not
what
you
want
and
at
one
extreme
we
can
see
ripple
aodv
play.
E
That's
the
pure
man,
a
very
challenging
environment
where
you
really
want
to
establish
a
little
route
that
you
can
use.
So
that's
a
very
extreme
use
case.
The
other
extreme
use
case
is
when
you,
you
want
a
much
better,
tighter
control
like
the
type
of
traffic
engineering
situation,
and
this
is
where
the
dow
projection
plays.
E
So
you
can
see
it
as
a
sdn
variation
of
ripple,
whereas
aodv
would
be
a
reactive
protocol,
adaptation
of
ripple,
all
of
them
play
with
the
same
messaging,
the
same
basic
structures,
but
they
they
operate
reactively
for
iodv
and
using
sdn
type
controller
based
routing
for
no
projection.
So,
that's
basically
how
I
would
position
it
now.
I
built
this
slide
the
sake
of
positioning
the
different
things
that
we
manipulate
with
this
document
and
how
it
it
builds
on
what
ripple
did
in
the
past.
E
E
What
the
first
type
is
what
I
would
call
a
segment
and
it's
a
serial
link,
it's
serial
set
of
hubs
and
that's
what
happens
with
a
project
down
which
is
sent
by
the
root
to
forwarding
node
g,
which
is
one
of
those
black
nodes
on
the
top,
and
the
the
p
down
will
follow
the
path
on
the
reverse
direction
of
the
packet,
which
is
true
for
redid
and
the
the
p
diode
contains
that
path.
So
it
will
basically
say
a
to
f
to
g.
It
is
sent
to
g.
E
C
E
E
E
E
They
are,
they
all
go
from
egress
to
egress,
right
in
grass
to
egress
it's
it's
a
path
that
you
establish
with
a
non-stirring
pidang,
and
it
it
pro
it
provides
a
loose
source
right
path
between
grass
and
grass,
and
you
can
see
the
first
leg
is
from
I
to
a
a
to
f,
f
to
g
g
to
e
and
and
through
that
you
can
reach
the
targets
ti's.
E
E
E
If
you
like,
see
that's
a
terminal,
but
it's
it's
more
complex
than
now,
because,
for
instance,
the
track
that
you
see
between
the
I
and
e
has
three
legs
in
it,
and
the
last
leg
is
I
to
b
to
e,
oh
by
the
way
I
say
I
to
b
to
e
I
so
that
I
should
have
said
I
to
a
to
e
b
four,
but
it
was
just
to
differentiate
that
I
was
going
via
the
top.
E
So
basically,
when
you,
when
you
establish
a
non-storing
loose
path,
you
just
provide
what
we
call
the
relay
in
this
picture,
which
are
the
nodes
which
are
which
appear
in
the
source
route,
so
the
source
route
says,
for
instance,
to
go.
I
to
a
to
e.
There
is
a
pedal
one
of
those
great
pedal
on
the
left,
which
is
set
by
the
root
to
increase
I
and
the
main
route,
the
root
of
the
meteor
deck,
and
it
says,
source
rod
path.
E
E
E
C
Sorry,
thank
you
so
from
subtracts
we
can
have
segments
and
legs
as
well
right.
E
E
It
goes
from
the
ingress
to
the
egress,
so
it's
only
it's
any
aggre,
it's
any
composition
of
legs,
so
so
the
the
track
between
a
and
b
is
not
a
subtract,
because
it's
it's
it's
what
we
say
in
the
language
is
we
place
the
bucket
on
one
subtract
and
what
it
really
means
is
you
place
it
on
one
more
legs
between
I
and
e
so
just
to
be
able
to
use
the
language
easily
right
now
the
text
uses
subtract
as
just
any
subset
of
the
legs
of
the
track,
and
that's
because
when
you
do
row
using
this,
you
will
not
place
the
packet
on
all
the
legs
in
parallel,
because
that
would
consume
too
much
energy.
E
E
So
so
that's
why
the
subtracts
are
any
subsets,
so
any
set,
which
is
a
subset
of
the
track
and
the
track
itself
is
the
set
of
legs.
E
So
this
is
ripple,
meaning
that
we
are
building
a
diode
between
I
and
e,
the
deodag
as
root,
which
is
I
it's
actually
reversible,
giving
you
a
deal
dag,
which
root
is
e
by
the
way
and
if
the
links
are
reversible,
the
other
will
be
if
the
links
are
not
reversible,
because
they
are
not
really
symmetrical,
then
you
use
any
in
one
direction
and
the
the
the
track
you
can
see
globally
from
the
external
view,
has
a
tunnel
between
I
and
e
and
effectively.
E
If
you
follow,
rfc
9008
you'll
find
that
the
targets
are
external
routes
based
on
all
destinations
for
this
geodec,
just
like
ripple
on
their
leaves,
if
you
like,
and
so
the
the
encapsulation
when
you
reach
ingress,
will
be
exactly
what
900
tells
you
so
the
increase
has
to
encapsulate
towards
the
egress,
which
is
the
sixth
hour
serving
the
600
watts.
Those
ripple
newer,
leaves
and
then
the
crest
encapsulates
and
and
and
delivers
to
a
target.
So
this
is
looking
from
a
distance.
E
I'm
also
summarizing
some
rules
of
how
all
this
work
works,
so
the
track
is
set
up
by
distilling,
legs
and
segments,
and
the
legs
go
from
the
ingress
to
egress
and
the
segments
basically
fill
the
gap
between
the
loose
hops.
So
the
segments
are
serial
hops.
E
E
E
The
fact
that
we
have
the
segments
so
we
have
the
storing
mode
that
enables
the
non-storing
to
be
loose.
Actually
it's
because
we
complete
the
gaps
with
the
story
that
we
can
have
non-storing
loops
and
in
particular,
we
can
leverage
that,
in
the
main
geodag
to
enable
non-storing
mode
ripple
to
have
loose
source
route.
E
So
you
basically
can
install
segments
in
the
main
geodag
so
that
your
packets
will
have
a
shorter
source
rotator
and
obviously
the
segments
are
also
used
inside
the
tracks
and
and
then
that
allows
you
to
again
to
specify
your
leg
as
a
loose
sequence
of
hops.
E
There
is
this
rule
in
the
document
which
we
could
discuss,
but
it
seemed
to
be
the
most
useful
way
of
doing
things.
The
tracking
grass
is
implicitly
a
target
when
you
do
no
starring
mode.
You
think
that
basically,
normally
you
the
stone
starting
mode
installs,
this
quarter
leg
this
tunnel
between
the
ingress
and
egress.
E
Okay,
okay
status
of
the
draft
and
the
other
rules-
you
can
read
them
so
we've
published
24
21
was
a
huge
review
by
taurus.
I
can
never
thank
you
enough,
but
we
discussed
that
the
atf
12
huge
restructuration,
because
taurus
3
came
from
outside
of
the
of
the
group,
and
so
he
asked
for
more
restructuration
to
help
the
first
time
reader
et
cetera,
et
cetera.
So
we
did
that
huge
restructuration
and
also
also
good
technical
points.
Then
we
had
michael's
review
and
then
again
it
went
kind
of
deep
terminology
clarification.
E
Also.
Mapping
to
that
net,
which
is
the
relay
nodes,
are
the
the
hops
the
loose
ups
and
the
forwarding
nodes.
That's
that
terminology
are
the
hubs
in
the
segments
one
in
one
out.
E
Then
we
have
23
with
liz
review
and,
like
I
said,
we
have
those
three
very
deep
reviews,
very
happy
with
them
and
lee
some
of
his.
His
questions
were
easy
to
answer
so
kind
of
did
it
with
version
23.
But
then
there
are
four
issues.
If
you
remember
that
we
discussed
at
an
interim
which
were
harder.
So
we
we
discussed
the
four
issues
on
the
list
and
I
published
basically
the
result
as
24..
E
So
my
my
view
is
that
24
fixes
all
the
the
kitab
issues.
One
of
them
was
how
many
targets
you
can
have
in
the
dao
and
lee
wanted
more
than
one.
But
how
can
the
controller
reach
more
than
one?
Maybe
it
can?
Maybe
it
cannot
maybe
the
path
that
would
reach
them
all
is
not
as
good
and
there
are
policies
in
play
and
we
did
not
want
to
enter
that
game.
E
So
what
I
said
is
okay
to
have
multiple
targets,
but
we
we
guarantee,
you
need
to
reach
the
first
one
in
order
and
then
the
the
controller
will
do
its
best
to
reach
also
the
others,
but
it
really
depends
on
what
can
do.
E
But
then
there
are
more
discussion
about
the
federation
flag
in
the
sau
is
a
link
between
two
siblings,
the
actual
or
not,
which
was
introduced
for
lease
review
on
23
and
and
the
text
was
improved
on
24..
E
Since
then
we
have
done
some
rephrasing
on
legs
and
subtracts,
which
is
basically
what
we
just
discussed
I've
I
had
to
publish
it.
I
could
publish
it
now
so
so
we
can
review
that
those
changes
they
are
minor,
but
it's
just
to
clarify
the
terminology.
So
we
can
really
speak
about
your
checks
and
we
were
waiting
for
him,
saris
review,
which
did
not
come.
E
I
mean
the
request
is
a
lot
of
work,
but
I
don't
think
blocks
were
good
plus
call,
because
we
had
all
those
three
reviews,
like
michael
tell
us
at
the
end,
so
that
was
since,
when
world
cup,
let's
go
start
and
everything,
I
would
hope
people
to
look
in
that
if
we
want
new
stages
to
be
more
specific
and
stuff
with
our
flows
that
they
have
in
mind
that
we
missed
in
particular
fields.
E
A
A
C
A
Okay,
I
just
the
first
time
with
my
with
the
phone
you
see,
so
there
you
go
okay,
so
this
document
has
been.
Did
we
even
get
as
far
as
working
group
last
call,
and
then
we
stopped
again.
I
can't
remember,
but
back
and
forth,
but
that's
the
story
so
far.
You
see
that
work.
A
There
we
go
so
we
got
quite
far
we
merged
in
another
document
with
a
different
metric
last
year,
and
we
had
a
lot
of
discussions
about
this
and
then
conrad
observed
that
this
new
extension
interacts
poorly
with
the
trickle
timer.
A
There's
a
word
for
spread
throughout
the
dodge
that
I
forget
it
will
propagate
through
the
dough,
dag
and
that
will
be
it.
If
the
root
wants
to
change
it,
it
would
need
to
you
know,
increment
the
dtsm
or
something
like
that
to
reset
the
trickle
timer
and
so
we
kind
of
reluct.
I
was
reluctant
with
that
change,
but
I'm
like
okay,
I
guess
that's
the
right
answer
it.
It's
a
compromise
and
rule
and
said
well
actually
that
just
doesn't
satisfy
his
need
for
it
all.
A
A
A
I
don't
feel
I
don't
feel
that
confident
about
the
whole
thing
until
I've
actually
written
some
code
about
it,
which
is
for
me,
won't
be
very
soon.
So
I
would
really
appreciate
some
more
deep
review,
particularly
if
you
have
writing
code.
I
don't
think
that
we'll
finish
the
the
this.
This
needs
some
effort
and
I
think
we've
talked
with
the
milestones
sometime
at
the
end
of
the
year.
This
will
be
ready
and
I
said
summer
I
think
that's
still
feasible.
A
That's
really
it
discussion.
If,
if
you,
if
you
have
questions
about
what
is
going
on,
I
have
some
other
slides
we
can
talk
to,
but
we
only
have
15
minutes
left.
So
I
don't
want
to
spend
a
lot
of
time
of
that.
If,
if
not
any
other
questions
from
the
group.
G
B
E
Yeah,
I
was
one
of
the
proponents
of
fixing
the
minimum
priority
at
the
root,
because
I
hope
I
demonstrated
on
the
mailing
list
through
a
number
of
arguments
that
it
just
could
not
work
the
way
it
was
specified
because
you
for
one
example,
you
would
get
different
values
from
different
parents
and
stuff
like
that,
and
also
there
was
the
policy
to
say
how
you
would
increment
it.
So
the
the
notes
could
increment
in
any
fashion,
and
that
was
really
weird.
I
think
the
the
lollipop
is
much
needed.
We
want
it.
E
E
It
doesn't
really
fix
the
prime
that
it
does
not
balance
the
dag,
but
another
argument
was
whatever
was
proposed:
there
did
not
fix
it
either
and
balancing
the
graph
is
a
very
difficult
problem,
but
it
won't
be
fixed
by
just
the
value
that
gets
incremented
down
the
geoduck
like
this.
E
If
you
look
at
it,
it's
a
global
problem
and
it
it
has
to
be
well.
I
looked
at
it
deeply.
The
only
way
I
found
to
fix
it
is
to
look
at
it
from
the
outside.
From
a
controller
perspective,
it's
it's
it's
a
very
hard,
prime
to
balance,
something
in
the
distributed
fashion.
Basically,.
A
So
there's
so
there's
the
thing
that
I
care
about,
which
is
a
is
a
less
pro
lower
property
or
than
balancing
the
graph,
which
is
that
nodes
don't
try
to
enroll
in
parts
of
the
graph
which
have
no
capacity
to
support
them.
So
I
I
think
that
that
that
it
does
solve
that
problem
right.
E
E
The
priority
into
the
beacon
now
so
so
it
could
be
done.
We
could
work
on
how
basically
someone
could
could
tell
that,
and
maybe
we
need
another
value
to
say
something
like
that.
Like
the
parent
would
say
below
me,
I
don't
want
anybody
or
something
but
th
this
will.
This
will
take
new
work
because,
for
instance,
do
you
want
nobody
because
of
amount
of
traffic?
Do
you
want,
and,
and
in
which
case
the
solution
could
be
slow
down
the
traffic?
Or
do
you
want
nobody,
because
you
don't
have
storage
mode
resources?
E
Well,
most
of
us
are
not
doing
story
mode,
or
do
you
want
to
do
it
because
you
don't
have
enough
done
story
mode
resources
where
non-storyboard
resources
are
really
to
your
children,
not
down
the
gear
deck,
so
you
don't
need
for
that
to
to
control
what
goes
beyond
you
in
the
dear
dark.
You
just
need
to
control
your
own
neighbor
cash,
and
and
for
that
your
beacon
is
enough.
A
E
And
for
the
for
the
children
of
yourself,
it's
not
your
concern,
because
that
does
not
affect
your
resources.
It
just
affects
the
support
that
you
get
through
and
now
the
question:
if
of
the
throughput
that
you
get
through,
is
yet
another
balancing
which
is
not
necessarily
related
to
the
children
right.
Some
children
can
speak
more
than
others,
etc.
So
it's
it's
it's
it's
a
lot
more
complex
than
than
than
what
the
draft
seem
to
enter,
and
and
so
let's
keep
the
draft
simple.
E
That's
what
he's
doing
well,
which
is
not
storing
if
you
haven't
been
cash,
prime
fix
your
beacon,
and
if
you
have
a
throughput
prime,
like
I
said
it's
probably
a
more
global
brand
than
this,
and
for
now
is
flood
control.
E
E
E
I
think
we
do
my
memory
of
that
is
all
right
is
what
conrad
said
is
probably
that
you
may
get
different.
A
E
If
we
use
the
traditional
lollipop
that
we
have
in
section
seven
of
65.50
yeah,.
A
E
We
need
a
eight
by
eight
bits
right,
but
do
we
want
that
or
just
do
we
just
want
a
a
rotation
of,
in
which
case
the
rotation
inside
is,
is
four
bits.
A
E
Specification,
you
should
have
one
bite
because
then
you
can
just
say
what
we
say
everywhere
else.
Just
look
at
section:
seven.
A
That's
what
I'll
propose
in
o5
then,
and
with
that
I
think
I'm
done.
Are
you
any
other
other
comments
from
other
people.
C
C
E
So
basically,
what
this
series
of
rfc
do
is
is
enabled
to
register
a
node
between
an
address
between
a
host
and
a
router
and
ask
the
router
to
to
redistribute
that
address
in
the
routing
protocol
of
the
day
that
can
be
vpn.
That
can
be
a
ripple
that
can
be
whatever
the
cool
thing
is
that
the
signaling
is
cryptically
abstract
to
the
routing
that
takes
place,
so
you
can
have
any
routing
protocol
or
both
it's
still,
a
simple
registration
with
the
same
format
exactly
by
the
host.
E
Well,
I
say
exactly
that
there
is
a
back
field
where
you
can
still
pass
some
hints
to
the
specific
routing
protocol.
If
you
need
to,
but
otherwise
it's
basically
a
routing
agnostic
registration
that
we
have
so
this
registration
was
initially
for
only
addresses,
but
we
are
now
with
this
draft.
We
are
proposing
to
to
extend
it
to
any
cast
and
multicast
address.
E
Yes,
and
that
was
pretty
much
a
mix
of
storing
mode
and
multicast,
but
we
we
don't
have
anything
for
non-storing
mode,
which
appears
to
be
the
most
important
case
for
the
actual
ripple
deployments.
So
part
of
what
this
thing
does
is
extend
855.
As
I
said
for
any
cast
and
multicast
which
which
the
sixth
floor
working
group
has
been
looking
at
and
so
far
I
mean
we
have
discussions
and
no
objections.
So
we,
I
think
we're
pretty
good
there
and
then
we
did.
E
The
ripple
counter
part
in
the
one
hand,
is
to
extend
the
reaper
and
wear
leaf
draft
rfc
now
to
inject
any
cast
and
multicast.
On
top
of.
A
E
But
also
extend
ripple
itself,
because
if
the
story,
the
geoduck,
is
operating
non-storing
mode,
we
could
effectively
leverage
that
as
well
for
multicast,
using
basically
a
distribution
set
of
set
of
tunnels
using
eight
zero,
zero,
eight,
but
across
a
collection
of
tunnels,
both
just
one
so
basically
distributing
a
crosstal
at
the
root,
which
is
what
we
are
proposing
with
this
map.
Five.
E
E
So
what
happened
to
this
draft
during
the
atf
112?
We
we
worked
on
cleaning
the
the
description
of
the
anycast
support
and
improving
the
backward
compatibility
pretty
much
with
in
1910
and
also
in
the
sixth
law
context.
The
eda
edac
exchange
had
to
carry
the
new
flags
that
were
introducing
to
say
any
customer
multiples.
E
Since
then,
we
we
made
we
bumped
the
version
to
from
two
to
four.
So
we
looked
at
some
concerns
some
feedback.
We
got
from
weissel,
actually
so
ysun
one
one
plans
to
use
this
work
and
they're
pressing
us
to
deliver
it.
E
One
issue
they
had
was
with
the
whole
node
multicastly
scope,
group
ffo2,
com,
com1,
and
they
did
not
want
the
nodes
to
have
to
register
to
it
because
in
normal
ipv6
this
is
implicit.
There
was
no
text
saying
yes
or
no,
so
we
clarified
that.
Yes,
you
don't
need.
You
do
not
need
to
register
to
this.
This
is
implicit.
E
Then
we
improve
the
description
of
how
address
protection
works
in
that
case
for
multicast
and
any
caskets,
and
we
also
align
the
description
with
the
work
which
is
proposed
at
best
to
redistribute
eight
five.
Four.
Five
and
now
this
new
work
multicast
and
any
cast
as
well
into
bgp
into
a
vpn,
so
the
workforce
for
this
group
is
manifold,
but
mostly
we
have
to
look
at
mob
five.
E
We
manage
the
collision
with
ripple,
aodv
right,
which
is
now
four,
so
we
we
have
five,
but
now
we
consider
whether
it's
okay
to
do
the
ingress
replication
has
proposed,
and
basically
the
draft
also
describes
how
you
do
you
select
one
of
the
possible
destinations
for
any
cast
and
with
this
I'm
out
of
my
time,
so
I
won't
describe
much
unless
you
tell
me,
I
have
more
time
yeah,
so
so
I'll
skip
on
this
slide,
which
gives
you
the
the
most
interesting
aspect
of
the
document
and
we'll
take
questions.
E
F
Sure
I'll
go
over
the
thumbnail,
so
I'm
going
to
delegate
that
to
our
chairs
to
go
talk
to
the
six
load
chairs
and
then
figure
out
figure
that
out
right
whether
we
have
two
documents
that
depends
on
each
other
or
one
point
to
the
other
or
whatever
that
is
or
we
work
on
one.
If
we
work
on
one,
we
need
to
figure
out
who's
going
to
last
call
it
who
owns
the
decision
et
cetera,
but
I
think
you
know
this
is
a
working
group
thing.
F
So
in
essen
dominica,
you
can
please
talk
to
the
six
law
chairs.
That
would
be
great.
I
will
give
eric
klein
a
heads
up
anyways
so
that
you
know
we're
all
aware
on.
What's
going
on
there.
You
have
talked
about
all
this
in
sixth
law
right.
E
E
F
Oh,
I
don't
have
a
preference.
No
again,
you
know
this
depends
on
what
is
going
to
go
into
each
of
the
two
documents
and
how
much
you
know
related
they
have
to
be
and
whether
they
are
two
at
the
same
time
or
two
at
different
times.
It
depends
on
the
implementation
and
other
statuses.
A
F
Right
so
again,
you
know
something
that
I
think
the
the
chairs,
in
that
case,
about
their
charter
with
eric.
You
need
to
figure
out
if
it's
in
charter
or
not,
you
know
we
have
all
kinds
of
documents
that
span
multiple
working
groups
all
the
time.
We
do
this
in
best
in
bgp
and
now
even
pim
and
other
working
groups.
F
Where
that
happens
all
the
time,
and
we
just
agree
on
how
the
document
is
going
to
be
processed
where
it's
you
know
usually
last
call
in
one
place
copied
to
the
other
one
where
we
need
to
get
consensus
from
both,
but
there's
a
set
of
chairs
that
are
going
to
own
this
so
yeah.
The
charter
issue
is
an
issue
I'll
I'll
talk
to
eric
about
that,
but
at
the
same
time,
I'll
keep
you
and
s
and
dominique
in
the
loop
on
that,
and
and
please
talk
to
the
six
door
chairs
about
it.
E
A
E
G
C
Okay,
thank
you
very
much.
We
will
discuss
with
the
six
law
chairs.
We
are
running
out
of
committee
minutes,
so
we
are
going
to
send
the
conclusions
in
the
mailing
list
and
trigger
the
corresponding
threads
and
the
meaning.
Some
other
commands.
C
There
is
a
comment
in
the
chat
atman
rashid
that
is
like
are
not
both
protocols
dependent
to
each
other.
I
mean
representing
japan,
carson
replace,
not
okay,
some
additional
common
questions,
michael
barger
issue,
a
red
issue
in
your
slide,
but
we
can
take
the
main
list
because
we
are
out
of
time.
C
Okay,
I
think
we
can
conclude
the
meeting
if
there
is
no
additional
comment.
Questions.