►
From YouTube: IETF-SCHC-20230613-1400
Description
SCHC meeting session at IETF
2023/06/13 1400
https://datatracker.ietf.org/meeting//proceedings/
A
Hello,
we,
as
customer
you
will
be
waiting
a
little
bit
until
more
people
join
and
then
we'll
start.
B
No
okay:
he
wants
to
to
have
a
see
the
location
after
the.
A
Okay,
that
will
be
discussed
in
the
bashing,
but
I
don't
see
why
we
could
not
do
that.
A
Okay,
I,
don't
see
that
Alex
is
coming,
so
I
will
start
doing
the
usually
tradition
slides.
So
this
is
a
negative
meeting.
The
usual
ITF
not
12
applies
in
particular
the
IPR
rule.
So
if
you're
aware
of
any
IPR
on
subject
being
disclosed
today,
please
announce
this
APR
or,
let
us
know,
let
us
know
after
the
meeting,
also
the
usual
anti-arrestment
and
code
of
conduct
through
all
supplies.
So
if
you're
not
aware
of
those,
please
refer
to
BCP
25
and
PCP
54.
A
and
apply
them,
and
you
may
always
report
to
the
embeds
team.
The
link
is
given
above
to,
if
you
have
any
concern
about
any
harassment
taking
place.
So
the
original
agenda
is,
as
this
and
castan
has
asked,
to
move
the
Errata
discussion
before
the
C
delegation.
A
I'm
hearing
nothing
saying
nothing
so
I
guess
we'll
we'll
move
forward
and
start
with
the
Errata
discussion.
A
So
I
now
raised
a
number
of
topics
that
the
discussion
has
started
remaining
list.
So
I
would
like
to
take
the
opportunity
of
this
call
to
to
get
some
resolution
some
some
of
the
items
that
announce
right.
So
we
can
input
text
in
the
human
architecture
and,
finally,
time
permitting
we'll
discuss
the
oam
draft
and
that
will
be
lojo.
A
So,
let's
move
on
because
we
have
a
huge
set
of
things
to
discuss
today.
A
So
as
far
as
as
it
goes
for
the
lp1
documents,
where
we're
done
all
the
mp1s
documents
that
were
active
but
then
P1
have
been
moved
to
Chic
and
the
one
which
were
in
the
isg
review
queue
I
have
now
been
handled
and
you
can
see
that
the
compound
deck
is
in
edit
State,
meaning
that
DRC
editor
is
actively
editing.
It
and
Six
Flags
was
missing
the
reference
to
that
one.
A
So
I
guess
when
the
the
sheet
company
is
complete,
both
shikovacic
flux
and
Shake
Kombat
will
fly
srfcs,
meaning
that,
as
far
as
far
as
that
P1
is
concerned,
we
can
declare
Victory
with
a
lot
of
good
documents
issued
and
I
would
say,
Grant
history.
For
this
working
mode,
so
now
we
move
to
to
check
and
we
already
have
a
huge
number
of
things
on
our
plate.
A
The
first
thing
was
the
adoption
for
the
Chic
over
PPP
and
I
realized
that
we
have
not
done
this
adoption
unless
I'm
missing
something
that
I
forgot.
So
we
really
need
to
do
the
call
for
adoption.
So
if
you
have
any
well,
it
will
be
discussed
on
the
next
slide.
But
please
think
if
you
are
against
calling
for
adoption
on
this
one,
because
we
discussed
that
otherwise
we'll
do
the
call
after
that
we.
A
Are
supposed
to
adopt
the
the
lp1
architecture
as
Chic
architecture,
I,
think
that
was
done
so
I,
don't
know
what
item
for
me
is
I
think
we're
all
set
there.
Maybe
it
was
just
making
sure
that
the
Sheikh
working
group
accepts
to
continue
on
the
mp1
architecture
document
I,
think
that
was
the
topic,
so
that
is
being
every
discussed
today.
A
Then
we
we
are
supposed
to
to
adopt
next
month
the
OM
documents-
that's
the
fourth
topic
that
will
be
discussed
today
in
time
permitting
and
then
we're
supposed
to
work
on
Bob's
draft.
But
Bob
is
not
with
us
and
started
a
bit
for
today,
but
we'll
discuss
it.
The
next
iatf,
so
in
more
detail,
the
Chicago
PPP
we
have
to
discuss
adoption
and
the
OEM
draft
I
mean
does
not
exist
either
on
lp1
or
or
check
working
group.
So
we
need
to
reboot
that
document.
C
A
C
A
C
Do
you,
okay,
so
I'm
going
to
do
that
and
and
then
we
hopefully
will
you
know
we'll
get
the
adoption.
So
it
was
two
weeks
right.
The.
C
A
D
Thank
you,
Pascal.
There
should
be
one
only
so
we
should
be
good
with
this.
Thank
you,
okay.
This
is
about
resuming
the
discussion
that
we
quickly
started
in
Yokohama
about
the
Arata
48824
and
something
actually
happened
since
then.
D
D
D
That's
the
first
one
that
we
quickly
discussed
is
specifically
about
the
section
on
the
options
size
once
I,
still
proxy
or
IM
proxy
scheme.
Discussing
the
case
where
CDA
is
value
sent
and
my
understanding
from
8724
and
8824
is
that
in
that
case,
TV
is
empty
or
undefined
ISO
both
used,
but
not
not
sent
really.
So
there
was
a
comment
from
one
about
the
the
choice
of
not
sent
instead
of
not
sent.
D
True
I've
said
that
in
both
Dimension
documents,
but
again
when
CDA
is
values
and
I
found
everywhere,
a
TV
empty
or
undefined
while
not
sent,
is
something
indeed
used,
but
for
other
fields.
So
the
proposal
would
be
to
consider
empty
here
too
consistently
with
other
sections.
D
D
Thanks
and
yeah,
the
text
in
the
new
draft,
referred
to
here
by
the
way,
is
in
fact
proposing
the
new
text
proposed
in
the
rata.
In
case
we
go
for
accept
and
offer
document
update,
run.
B
D
A
Okay,
so
for
the
minute
do
we
agree
that,
basically,
the
group
agrees
with
what
Marco
is
addressing
I
mean?
That's
that's
what
you
were
saying
in
the
whole
right.
We
are
all
in
agreement
with
Marco
situation
here,
right.
A
F
G
E
C
From
Ivan,
who
was
I
just
have
a
question
about
young
data
model
is:
is
there
an
empty
type?
How
is
it
defined
in
9363,
so.
B
F
Yeah
I
do
understand:
I
wasn't
referring
to
the
field,
that's
just
the
TV
itself.
I
mean
it's
a
it's
a
construct
right
when
you're
defining
a
TV,
you
have
to
to
provide
some
kind
of
way
of
saying
the
the
TV
is
of
a
given
size.
It
makes
sense
that,
for
example,
if
you're
providing
a
string
value,
there
is
a
backslash
zero
that
is
implied,
but
yeah
I
mean.
F
G
C
Length
zero
and
there
is
a
difference
between
a
null
string
and
an
empty
string
and,
in
any
case,
I
think
that
here
it's
pretty
clear
that
if
you
have
the
ammo
set
to
ignore
and
the
CDA
sent
to
Value
sent,
then
the
TV.
Basically
you
don't
care,
but
in
order
for
us
to
be
complete,
we
say:
okay.
Well,
we
put
some
value
some
special
value
to
to
indicate
that
you
know
that
there
is
no
target
value
there.
C
So
The
Proposal
is
to
put
an
empty
TV.
Isn't
that
more
ambiguous
than
than
saying
that
there's
a
special
type
of
Target
value,
which
is
called
empty
I,
don't
know.
C
F
Yeah,
you
made
a
good
example
with
the
with
the
string
the
empty
string,
because
the
empty
string
is
elf
hands,
one
one
byte
and
there's
the
backslash
zero.
So
it's
empty
for
a
string.
The
the
length
of
the
field
itself
is
going
to
be
of
size.
One.
A
C
So
so
I
mean
yeah.
The
point
is
that
empty
can
be
ambiguous
depending
on
the
in
some
programming
languages.
It
can
mean
something
and
in
some
others
like
well,
it
can
be
this
or
that
python
is
very,
you
know
it
can
have
empty
to
be
empty
string
or
no
I
mean,
maybe
not
no,
but
you
know
you
can
have
a
couple
of
things
that
resolve
to
empty.
In
any
case
like
it,
it
seems
that
and
in
this
problem
the
question
to
Marco.
C
So
how
so
here
the
erratum
is
because
it
is
it.
You
say
that
it's
like
not
sent
so
here's
the
way
I
an
interpret
not
sent
is
like.
Okay,
there
is
a
special
value
that
indicates
that
you
know
that
Target
value
is
not
sent
right.
So
is
it
a
good
choice
to
have
put
it
the
same
thing
as
the
CDA?
Then
you
know
that's
a
different
story,
but
here
is
like
a
special
value
like
no.
No,
it's
a
special
value
and
well
there
were.
C
C
So
here
the
point
is:
are
we
do
we
want
to
define
a
new,
empty
type
which
says
hey?
Well,
that
or
should
we
say
well,
the
TV
should
be
ignored.
D
D
B
B
And-
and
it
depends
it's
in
the
world,
because
you
cannot
make
a
hole
on
the
paper
when
you
yeah
you
do
it,
but
in
in
the
young
data
model
it
doesn't
exist
and,
for
example,
in
open
cheek
it
can
be
viewed
as
none,
but
that's
an
implementation
issue.
E
I
think
that
Marco
rice.
E
E
D
So
probably
not
set
it's
good
to
use
those
in
some
of
those
sections
put
it
there
I'm
pretty
sure
it's
not
so.
Another
point
is
consistency.
Of
course
this
was
the
only
outlier
I
could
find,
but
if
it's
fixed,
depending
on
the
fix
we
choose,
the
fix
has
to
be
propagated.
C
C
Because
it
avoids
adding
a
new
special,
constant
right,
but
I
mean
that
goes
I,
think
that
goes
beyond
a
technical
atom.
So
probably
yes,
as
Marco
and
Anna
said,
the
best
thing
is
just
settle
on
one
thing
and
and
and
make
sure
that
you
know
they
are
harmonious
right.
So
if
it's
the
setting
the
TV.
D
B
G
D
Okay,
moving
to
the
next
one,
then
yeah.
This
was
also
quickly
discussed
in
Yokohama.
It
was
about
the
case
when
observe
is
used
and
I'm.
Sorry
that
the
new
text
I
proposed,
tried
to
be
simple,
but
yeah,
distracted
from
the
main
point
and
opening
discussion
on
option
versus
extension.
That's
really
not
what
I
meant!
What
what
read
weird
to
me
was
that
the
co-op
option
seems
like
can
use
a
message
while
the
user
here
and
is
the
client.
So
maybe
it's
better
for
for
clarity
to
avoid
misunderstanding
to
just
consider.
D
Instead,
the
rephrasing
I
proposed
here
at
the
very
end
of
the
slide,
where
it's
not
about
option
altogether,
it's
more
about
describing
the
situation
and
and
who
the
real
user
is.
D
E
Yeah
I
think
that
it
was
very
confusing
because
we
are
really
compressing
the
option,
even
if
it's
an
extension,
but
when
we
make
Chic,
we
compress
the
option.
So
what
you
are
suggesting
is
more
clear
for
me.
D
And
now
so,
this
ended
up
easy,
94
right.
This
was
editorial
to
me
at
least
the
way
I
found
it.
It
really
looked
like
a
typo
that
that
digit
one
in
the
mo
field.
Now
it's
smart
technical.
D
If
I
understand
correctly,
she
is
proposing
to
fix
equal
one
to
be
equal,
but
then
to
replace
a
path
with
LM.
Basically,
but
Anna
at
least
I
mean.
E
Again,
I
think
you
find
another
Niche
for
the
very
old
version
of
the
of
this
document,
because
when
I
read
your
the
their
atom
I
say
what
is
this
equal
one
and
so
I
start
searching
where
we
put
this
equal
one
and
I
found
it
in
I.
Think
it's
it's
in
version
two
or
three
of
the
draft
and
we
never
correct
it
until
the
end.
Until
now
and
I
say
it's
not
possible,
we
cannot
have
a
m
matching
operator
equal
one
that
doesn't
exist.
So
you
are
right
is
an
error.
E
For
me,
the
correction
will
be
to
to
have
matching
operator
equal
and
the
element
in
the
Target
value,
because
path
can
have
different
elements,
so
during
pad
is
divided
in
in
different
elements.
So
in
Target
value
you
have
the
element
of
the
path.
D
D
I
don't
know
and
well
replacing
path
with
LM
is.
D
B
Yes,
it's
it's
an
historical
point,
so
I
think
we
I'm
about
to
discuss
that
much.
We
just
remove
equivalent
and
that's
all
I
think
what
I
now
wanted
is
just
path
is
Slash,
something
slash
something
slash.
Something
as
element
is
just
something,
so
it
may
be
very
clear,
but
I
think
it's
detailed.
E
E
D
Then
I'm
on
25
yeah,
we
didn't
see
this
this.
This
was
about
a
set
of
rules
on
compressing
Co-op
type.
D
Yeah
I
just
noted
the
sentence
in
the
text
that
would
really
suggesting
something
consistent
with
values
and
in
both
Direction.
Well,
you
actually
have
a
case
where
you
don't
send
nothing
at
all
in
a
certain
direction.
If
type
is
specifically
confirmable,
so
I
was
proposing
to
expand
the
text,
I
believe
to
describe
the
rules
in
the
table
in
a
clearer
way,
as
Anna
said
totally,
the
whole
rule
covers
both
type
and
code
separately
and
in
a
good
way.
D
The
text
was
just
about
not
suggesting
the
type
is
always
sent
because
there's
a
case
where
you
don't
send
it.
D
Thanks
then,
I
move
on
okay.
We
haven't
really
seen
this
before
I
scores,
all
three
together,
the
first
two
ones
were
originally
editorial.
Now
they
are
technical.
The
first
one
in
the
old
text
suggested
to
me
about
compressing
the
client.
Of
course.
The
point
is
about
compressing
the
request,
so
I'm
I'm,
just
rephrasing
here
to
make
that
clear.
E
D
I
think
they
were
mostly
looking
for
Eric
to
check
and
comment,
so
he
will
probably
object
if
editorial
is
more
appropriate.
E
D
Okay
yeah,
the
second
one
was
about
splitting
clearly
core
options
in
general
as
including,
specifically,
the
Oscar
option
indicating
the
use
of
all
score,
so
options
should
be
should
be
in
the
plural.
That
was
the
the
essence
of
the
ratan.
To
me,
that
was
editorial
as
well.
I,
don't
know
why
they
changed
it.
E
E
D
But
yeah,
especially
these
two
I,
think
I
feel
that
mostly
editorial
the
third
one
here
instead
was
technical
all
along
and
it
was
about
clarifying
the
the
behavior
of
some
co-op
options
per
Co-op.
So
your
iPath
is
not
mandatory
to
have
in
the
request.
As
the
old
text
says,
there
are
cases
where
it
can
be
absent,
while
on
the
other
hand,
it's
it's
not
really
supposed
to
be
in
responses.
So
the
new
text
should
clarify
this.
D
D
A
If
I
read
well
pretty
much,
everything
is
accepted
right.
It's
just
that
the
first
one
I
guess
the
conclusion
was
to
to
say
not
set.
Instead
of
not
sell
for
the
TV
right,
we
just
so
describing
an
empty
TV
is
not
what
we
want.
We
want
to
say
setting
the
TV
to
not
set.
Is
it?
Is
it
correct
enough?
Yes,.
E
Yes,
another
issue
there
I,
don't
know
if
it's
in
the
same
document,
you
add
more
things
than
the
rata.
Isn't
it
but.
D
A
new
draft
right,
the
new
draft,
the
new
draft,
as
of
now
was
intended
to
address
only
this
particular
rata,
basically
restating
that
small
section
on
those
options
with
the
new
phrasing
that
we
can,
of
course,
change
using
unsetting.
But
then
maybe
there
is
more
text
from
ATA
24
to
be
fixed.
With
this
new
terminology,
we
can
also
do.
A
D
A
So
so
we
I
will
talk
to
Eric
about
saying
that
we
I
actually
can
basically
well
it's
the
one
we
discussed.
But
let's
see,
if
you
can
just
give
me
this
list
of
ideas
and
so
I
will
I
will
post
a
ring
that
we
accept
them
just
the
7391.
Please
rephrase
it
before
we
say
we
accept.
A
D
A
You
and
next
is
the
seed
I
guess
we
are
already
late
to
home,
cover
all
our
topics.
So
let
me
share
this.
A
You
want
to
do
it.
I
do.
B
A
B
So
it's
okay!
It's!
Okay!
Do
you
see
the
slides?
Yes?
Yes,
yes,
okay!
So
it's
a
presentation
we
should
have
done
in
in
Yokohama
and
I
didn't
find
the
real
PDF
version,
so
I
took
back
the
one
from
Yokohama.
That's
why
you
have
the
this
number,
but
nothing
has
changed
and
I
had
to
refresh
myself
about
what
we
have
to
to
say
on
it.
B
But
in
fact
it's
to
talk
about
the
process
to
do
see
the
location,
because
there
is
a
process
that
is
defined
by
the
the
core
working
group,
but
is
fine,
but
I
think
we
can
go
a
little
bit
more
deeper
to
optimize
things,
especially
for
for
Chic,
where
we
need
to
have
very
constraint,
representation
of
the
young
databoder
so
just
to
refresh
how
it
works.
B
So
on
the
left
and
sides
you
have,
for
example,
an
RFC
that
will
find
a
young
data
model
and
inside
you
will
have
some
identifier
we
should
in
in
Chic
to
use
identifier,
because
this
way
we
have
a
unique
way
to
represent
something,
and
anybody
can
also
add
a
new
your
new
identifier,
and
we
have,
of
course,
the
data
model
by
itself
that
is
represented
by
a
structure
and
in
core
conf.
This
structure
is
Delta
encoded.
B
It
means
that
you
don't
put
always
the
c
number,
but
you
put
a
Delta
regarding
the
the
previous
one,
so
that
sign
it
works
well.
But
in
our
vision,
where
we
want
to
augment
a
model,
if
we
do
another
NFC
with
another
young
data
model,
then
it
will
have
the
same
structure
so
with
identity
at
the
beginning
and
then
the
leaf.
B
What
we
we
had
and
since
there
is
several
rules
that
say
that
you
need
to
use
at
least
50
seeds
when
you
ask
for
a
location
and
that
you
have
identity
that
are
always
at
the
top.
It
means
that
for
the
Delta
encoded,
if
you
in
a
structure
you
take
leaves
from
one
young
data
model
and
over
and
it's
data
encoded,
so
the
Delta
can
be
higher
and
of
course,
if
it's
only
the
difference
is
only
between
-24
and
23.
B
B
So
what
we
so
we
make
some
measurement.
So
first
hypothesis
is
that
we
have
a
Chic
model
that
has
been
defined
in
RFC
9363
and
it
will
be-
and
we
will
add
new
things
around,
and
so
we
make
some
different
that
we
did
the
test
to
see
what
will
be
the
size
of
this
allocation.
B
So
here
is
a
proposal
we
come
back
after
on
that,
but
especially
here
you
have
a
new
allocation
of
that
could
be
done
for
the
group.
It
means
that
here
we
have.
So
if
you
look
at
what
will
be
done
by
default
but
pyong,
it
means
that
you
will
take,
for
example,
5000,
and
then
you
will
put
your
first
element
secondary
Max,
cetera
Xterra
until
until
the
end.
B
So
here
what
we
propose
is
to
have
a
more
tricky
way
to
to
do
things
and,
for
example,
here
you
have
one
main,
the
top
of
the
structure
that
will
be
put
in
position
5024,
which
means
that
this
way
we
have
24
value
before,
where
we
have
the
small
of
Delta,
where
we
can
put
things
on
23
value
after
where
we
can
also
put
things,
and
this
way
we
we
have
free
space
for
augmentation.
B
Of
course,
this
value
over
overlapping.
So,
for
example,
in
blue,
you
have
four,
the
ATF
Chic
and
in
green.
You
have
for
another
element,
especially
same
thing
in
in
yellow,
so
here
we
have
a
test
of
allocation
that
should
be
discussed
after
in
the
group.
But
for
example,
here
you
have,
if
you
look
at
the
value,
500
5070.
So
here
you
have
an
entry,
so
that
describe
rulant
tree
and
we
put
all
the
element
that
we
expect
just
after.
B
We
have
free
space
and,
after
that
we
have
over
element
for
fragmentation
that
will
here
it's
in
red
on
the
left
column,
because
here
we
have
a
coding
on
two
bytes,
but
these
are
less
frequent
elements.
So
what
we
will
have
to
discuss
if
this
kind
of
proposal
is
accepted
is
how
we
place
all
the
identifier
we
our
elements
we
have
in
in
a
Chic
data
model
to
be
sure
that
we
can
extend
easily
things
with
new
Option
and
also
that
the
size
of
the
representation
can
be
very
compact.
B
B
B
We
have
also
vom
model,
that
is
in
draft
mode,
where
you
have
10
identity,
mainly
and
one
data,
and
we
have
access
control.
That
also
add
four
elements
in
the
structure
for
the,
of
course,
this
is
just
example.
We
are
working
on
the
access
control
run.
We
will
have
I
think
a
different
structure,
but
here
what
is
important
is
how
we
represent
this
this
information.
B
So
let's
have
a
look,
so
we
make
some
measurement,
so
we
have
three
type
of
rules
so
set
of
rule
9363,
which
is
the
basic
one.
So
with
two
is
all
that
comply
with
the
RFC.
We
have
set
of
rule
aom,
and
so
we
add
a
new
identity
to
the
model
and
a
set
of
full
RM,
plus
Access
Control,
where
we
add
new
identities,
but
also
new
leads
in
in
the
structure.
B
So
the
First
Column
is,
for
example,
if
we
take
seeds
in
the
public
range.
So
in
the
million
range-
and
so
here
we
see
the
size
of
the
the
structure
for
different
pool,
but
are
listed
in
the
draft.
So
it's
2194
extra,
so
you
see
the
augmentation
because
we
add
new
identity
or
we
add
new
new
list
now.
The
good
news
is
that
if
we
take
the
regular
way
to
do
via
location,
it
means
that
we
took
take
the
ietf
space,
but
we
separate
all
the
data
model.
B
Then
we
have
a
reduction
of
at
least
400
bytes
or
more
to
with
this,
so
it's
better
than
using
another
space.
So
that's
why
it's
important
for
for
the
group
to
request
some
some
value
now,
if
we
have
ITF
Group,
which
means
that
we
are
using
the
allocation.
I
show
you.
So
we
have
a
diminution
of
19
bytes.
In
the
first
case,
we
have
24
2,
but
here
it's
it's
logical,
because
we
hear
the
differences,
because
we
have
a
clever
allocation
and
we
have
smaller
Delta.
But
the
difference
is
not
that
I.
B
But
when
we
look
at
the
last
element,
so
the
store
iom
plus
access
control,
then
here
we
introduce
new
leads,
and
here
we
see
that
we
have
a
better
reduction
of
50
bytes
and
then
we
have
the
last
colon
or
the
Pentium
column,
where,
if
it's
called
ASG
dream,
it
means
that
we
take
tens
value
in
the
smallest
value
possible
and
that
are
reserved
by
AIG.
And
then
we
have
to
discuss
and
prove.
That
is
interesting.
B
So
here
we
take
only
one
percent
of
the
space
and
you
see
that
and
we
allocate
them
to
the
most
common
element
like
for
matching
operator
and
CDA,
and
here
we
have
a
very
strong
reduction
in
size
of
the
representation.
And
in
fact
you,
if
you
look
it's
about
10
percent
of
the
representation
in
inches,
so
that's
some
some
measurements.
B
So
for
me,
what
we
can
do
in
the
group
is
that,
instead
of
allocating
a
seed
Ranch
for
each
RFC,
I
think
the
working
group
should
ask
for
a
pool
of
addresses,
so
I
put
500
because
I
think
it's
a
reasonable,
reasonable
number.
But
we
can
maybe
go
to
1000,
maybe
ask
for
isg
for
the
most
common
values,
to
use
very,
very
small
seeds
for
this
identity
and
each
time
the
working
group
issue
a
new
RFC
containing
a
young
data
model.
B
Then
we
have
to
do
this
manual
location,
which
is
not
so
complex
to
do,
because
we
can
do
use
pyong
to
do
an
allocation
with
dummy
numbers,
and
then
we
can
have
a
CSV
file
as
custom
proposal
and
we
can
change
this
w
value
to
the
value
that
has
been
allocated
by
by
the
group.
So
I
don't
know
if
I
was
clear
in
this
short
presentation.
But
if
you
have
questions
no
comment.
G
Yeah,
the
reason
why
I
wanted
to
be
part
of
this
discussion
is
that
this,
of
course,
has
has
an
influence
on
what
we
actually
write
in
the
course
it
document
and
right
now
the
corset
document
essentially
States.
A
few
requirements
on
how
to
set
allocation
should
be
done,
but
then
goes
on
to
to
derive
a
procedure.
How
to
do
this?
That,
maybe
is
too
narrow
and
I,
think
we.
We
should
spend
a
little
time
making
sure
that
the
statement
of
the
requirements
is
complete.
B
Okay,
I
think
in
the
far
away,
but
in
in
the
core
confer
you
have
a
way
to
ask
for
a
value.
A
working
group
can
ask
for
for
Value
right
now.
G
I
think
we
we
really
have
to
read
carefully
what
what
cause.
It
is
really
saying
about
this,
because
okay,
yeah
I
did
saying
a
lot
and
it's
giving
a
really
nice
default
procedure
that
that
people
can
be
using.
But
we
have
to
be
careful
to
say
what
part
of
that
procedure
is
actually
a
requirement
and
what
part
is
just
the
way
we
recommend
people
to
to
do
if
they
don't
know
a
reason
to
do
something.
More
elaborate,
elaborate
like
what
we
are
proposing
here.
B
G
G
So
I
really
like
the
the
approach
for
doing
these.
These
allocations
in
a
space
where,
where
space
really
matters,
so
this
is
a
good
work.
We
just
have
to
make
sure
that
that,
because
it
properly
recognizes
these
cases
and
and
describes
its
own
procedure,
just
as
as
the
default
way
of
doing
things.
C
So
so,
just
one
question
here
on
on
the
next
steps
on
this
document
and
thank
you
very
much
for
the
presentation
and
and
custom
for
the
for
for
the
very
good
observation
I'm
looking
for
for
taking
part
of
the
discussion
and
for
making
sure
that
the
course
seed
document
is
is
aligned.
C
G
B
A
A
B
B
I
think
it's
it's.
It
is
now
being
specified
by
commikor
or
comic.
G
Yeah,
so
basically,
we
don't
want
to
task
Ayanna
with
doing
something
that
requires
a
deep
knowledge
of
the
subject
space,
so
they
are
starting
to
do
with
do
this
with
Yang
in
general,
but
I
think
before
we
get
them
to
actually
do
to
sit
allocations
as
a
service
that
will
take
some
time.
So
the
the
current
course
hit
document
essentially
replaces
the
owners
on
the
designated
expert.
G
G
Well,
the
the
isg
values
is3
value,
so
they
probably
won't
allocate
these
unless
there
is
a
document
that
justifies
that.
G
A
Okay,
so
we
are
very
close
to
the
top.
We
won't
have
time
to
enter
the
discussions
and
the
architecture.
So
what
I
suggest,
if
you
don't
mind,
is
we
discussed
a
little
bit
what
you
want
to
do
with
the
oam
draft,
so
you
we
just
have
five
minutes.
If
it's
okay,
so
you
can
tell
us
what
is
the
future
for
this
document.
B
Sorry
for
viewers,
so
I
have
to
find
it.
I,
don't
see
it
in.
Oh,
yes,
it's
here,
okay,
so
so
it's
a
document
that
we
have
done
with
Dominic,
and
so
we
didn't
publish
right
now,
the
zrfc,
because
it's
we
have
some
issue
with
XML,
but
you
can
find
the
document
and
the
lp1
GitHub.
B
So
if
you
want
to
have
access
to
to
it,
so
I
will
say
quickly
that
we
focus
only
on
icmp
messages
that
are
defined
in
the
RFC
4443
and
to
do
the
the
compression
that
is
not
just
a
document
where
we
Define
the
rule
for
compression
like
for
Co-Op.
But
in
fact
we
add
also
a
new
feature
to
Chic
that
are
I,
think
very
important,
also
for
the
evolution
of
of
chic.
So
the
different
element
we
we
have
is
first
the
proxy
that
will
process
directly.
The
information
I
will
show
you
some
examples.
B
We
have
also
icmp
generation
that
can
be
done
for
when
we
haven't,
we
find
an
error
in
the
compression
or
fragmentation,
and
also
we
propose
new
CDA
or
new
matching
operator
for
for
compressing
inside
a
message
so
and
everything
has
been
implemented
in
opencheck,
so
we
can
test
it
and
I
can
show
you
how
we,
maybe
during
the
academy
we
I,
can
show
if
people
are
interested
foreign,
so
here
just
quickly
the
different
elements.
B
So
what
is
new
is
that
now,
of
course,
you
can
compress
packet
on
the
link,
but
here
we
have
a
proxy
that
allows
you
to
answer
to
the
Peaks,
so
the
pink.
So
in
the
middle
here
you
have
the
core
Chic,
and
so,
if
a
Chic
message
has
been
received
by
the
kaushik,
it
will
answer
two
Echo
requests
directly
and
it
will
not
be
sent
to
the
device,
but
it
is
specified
for
a
certain
period
of
time
on
when
this
period
of
time
have
expired.
So
the
core
Chic
will
not
answer
to
a
request
anymore.
B
A
B
It's
good
for
the
moment.
I
think
it
will
need
to
have
another
action,
because
the
current
action
is
to
do
this,
and
but.
B
A
So,
okay.
B
So,
but
yes
so
era,
it's
so
what
we
have
done
here
is,
for
example,
you
have
a
device
that
sends
a
message
and
the
message
generate
an
icmp
message
somewhere
on
the
internet,
so
the
corsic
receive
an
icmp
message,
and
here
what
we
propose
is
to
add
the
to
recompress
the
message
that
is
Echo
by
scmp
and
to
send
to
the
to
the
device
the
error,
icmp
message
that
is
compressed
on
the
payload.
That
is
also
compressed
using
the
rule
that
the
Corsica.
B
So
this
means
that
we
have
two
new
ma
that
say:
match
Rule
and
much
or
reverse
match
rule.
That
means
that
we
are
using
matching
in
the
same
direction
or
in
the
opposite
direction
and
same
thing.
We
can
send
a
compressed
things
or
reverse
compress.
A
A
I
can
suggest
you
know
there
is
very
little
traffic
coming
from
the
device
and
what
I'm
really
concerned
is
if,
because
we
don't
have
I
mean
that's
part
of
the
architectural
discussions,
but
we
don't
have
an
order
for
the
rules,
so
you
never
know
if
you
will
compress
it
the
same
way
as
it
was
comprised.
So
what
you're
setting
back
is
probably
not
the
same
bits
as.
E
B
And
the
last
last
thing
is
that
we
introduce
also
some
icmp
message:
error
that
can
be
generated
by
the
kaushik.
For
example,
we
can
use
type
zero,
so
it's
error
code,
three.
It
means
that
us
not
found
it
means
that
a
package
arrives
to
the
core
Sheikh
and
it
didn't
find
any
rule
for
any
devices.
B
So
here
we
send
this
message
if
the
device
is
found,
but
there
is
no
matching
rule,
we
say
that
the
port
is
not
found
and
I
think
we
don't
have
to
introduce
new
icmp
messages
because
it
can
be
a
security
issue.
For
example
to
say:
oh
I
am
a
poor
device
that
is
on
the
lp1
network.
So
don't
attack
me
and
here
it's
regular
messages.
So
it's
better
and
another
one
that
has
not
been
put
in
the
draft
right
now.
B
But
when
you
receive
a
too
big
packet
and
your
fragmentation
rules
cannot
carry
this
large
packet,
then
we
can
send
an
ICM
pip
to
Big
message
and
so
I
just
put
here
some
example
that
is
derived
from
open
cheek
and
in
the
second
column
on
the
bottom.
You
see
the
icmpv6
payload,
and
here
we
have
the
matching
operator
that
is
used
and
we
send
nothing,
but
it
means
that
if
the
matching
upper,
if
we
don't
find
any
reverse
rules,
then
we
will
not
apply
this
Rule
and
the
last
column.