►
From YouTube: IETF-CORE-20210915-1400
Description
CORE meeting session at IETF
2021/09/15 1400
https://datatracker.ietf.org/meeting//proceedings/
B
A
Yeah,
we
know
what
to
say
most
real
content
will
come
from
carsten
today.
I
think.
C
A
C
E
A
Okay,
but
we
are
for
past,
so
let's
start
the
meeting
now
so
welcome
everybody
to
this
co-working
group
interview
meeting.
This
is
marco
tiloca.
My
co-chair
is
how
many
minutes,
and
of
course
this
is
an
official
idf
meeting.
So
the
not
well
applies
be
sure
to
get
familiar
with
that,
if
you're
not
already,
it's
not
just
about
ipr,
it's
also
about
our
code
of
conduct
so
be
nice
and
professional
to
each
other,
and
we
need
some
help
with
taking
the
minutes.
A
That
shares
can
take
notes,
of
course,
but
please
volunteer
to
take
minutes
in
the
academy.
A
So
just
not
before
we
go
through
the
agenda.
We
are
going
to
have
four
interim
meetings,
including
this
one
before
itf-112
and
the
schedule
is
regular,
so
every
other
wednesday
starting
today
same
time
for
each
meeting
and
it's
alternating
with
seaboard
as
usual.
A
The
agenda
for
today's
especially
status
update
for
a
number
of
working
group
documents,
I
believe
cars,
then
we
will
need
most
of
those.
So
it's
nml
data
ct
the
com,
the
core
conf
documents
and
href
and
coral
works.
A
Then
we
can
have
a
discussion
on
the
new
proposed
charter
text
that
we
circulated
last
week
on
the
mailing
list
and
in
the
code,
md
notes.
And
finally
I
will
cover
a
new
draft
related
to
oscore.
That
was
in
the
agenda
for
the
july
atf
meeting
and
we
had
to
skip
due
to
lack
of
time.
B
I
just
wanted
to
see
if
we
can
have
some
update
for
the
documents
that
are
still
in
isg
evaluation,
so
the
core
sid
and
the
young
seabor
documents.
I
marked
these
revised
that
he
needed
some
months
ago.
B
I
don't
know
if
it's
like
57
days
or
it's
57
days
inside
his
job
mission,
but
they've
been
in
there
for
a
while
and
they're,
currently
blocked
with
four
passes
on
the
seed
document
and
two
discusses
on
the
young
seaboard,
and
so
I
just
wanted
to
know
if,
if,
if
the
authors
have
implemented
some
changes
in
the
github
or
if
they're,
waiting
on
80s,
if
I
should
do
anything
about
those
or
if
you're,
waiting
on
the
authors.
D
Well,
that's
on
the
agenda,
but
you
can
answer
now
that
we
have
pulled
it
up.
So
that's
that's
of
the
two
for
coconut
documents
listed
here.
It's
the
first
two
young
zebra
and
sid,
and
there
were
some
some
really
good
comments,
but
unfortunately
comments
that
actually
require
some
some
thinking
how
to
properly
handle
them
and
I'm
I
have
been
for
57
days
trying
to
find
a
day
where
I
can
do
that
that
they
might
be
tomorrow.
D
But
I
already
thought
that
a
few
times
so
yeah
and
that's
a
bit
of
a
problem
for
me
right
now.
B
A
D
D
Generally
have
gone
to
the
core
mailing
list.
So
if
you
look
for
something
like
discuss
on
draft
itf
call
yang,
then
you
will
find
some
things
or
no
objection
on
or
maybe
just
on,
undrafted
call
yang.
Then
you
will
find
them.
B
Might
not
get
an
email,
so
it's
always
good
to
check
check
the
ballot.
So
in
this
case
everything
was
sent
by
email,
so
it
doesn't
apply
to
this
one,
but
for
other
documents.
B
I
I
kind
of
remember
some
discussion
in
the
mailing
list.
That's
why
I
was
wondering
if
there
was
anything
on
my
side
that
needed
to
be
done,
but
I
also
think
that
the
discussion
didn't
conclude
and
so.
A
Okay,
thanks:
we
can
jump
into
the
agenda
and
well
kirsten
it's
up
to
you
for
sending
ml
data.
Cts.
First
item.
D
Yeah,
so
we
had
two
reviews
so
far:
francesca's
and
and
the
gennad
review,
which
was
pretty
nice
from
christopher
heimberg,
so
francesca's
review
was
really
pushing
the
document
forward.
Thank
you,
and
just
also
had
some
some
useful
comments
and
I
have
processed
these
comments
with
ari
and
we
now
have
a
dash
of
five,
which
I
think
is
covering
these.
C
Two
just
a
moment:
carson
marco:
are
you
sharing
these
lights
or
is
carsten
showing
them.
D
B
D
D
Yeah,
so
I
already
said
where
the
first
two
documents
are
stuck
right
now
and
the
the
other
two
documents
essentially
have
one
big
question
for
comai:
that
we
have
to
answer,
which
is
the
the
the
lack
of
commonality
between
the
encoding
of
the
key
parameters
in
the
uris,
and
we
still
haven't
found
out
why
we
are
doing
this
in
in
this
little
bit
weird
way,
so
that
that's
something
I
would
like
to
to
find
out.
D
D
So
this
this
is
the
the
dash
comma
thing
and
yang
library
really.
Somebody
has
to
sit
down
and
and
track
everything
that
has
happened
in
the
young
universe,
since
that
was
done,
because
it's
essentially
a
constrained
version
of
of
a
document
that
essentially
has
has
received
a
lot
of
attention
in
in
the
young
world.
So
this
this
is
a
nice
little
exercise
for
a
student.
Unfortunately,
I
don't
have
a
student
working
on
that
certificate.
At
the
moment,.
D
D
A
A
D
Yeah,
so
the
I
think
we
have
talked
about
href
and
right
now.
The
situation
is
that
we
we
have
a
draft
that
pretty
much
works.
D
We
still
have
limited
implementation
experience,
so
I
have
a
half
implementation.
Christian
has
a
complete
implementation,
but
we
probably
want
to
collect
more
more
experience
with
that
and
last
week
in
the
cbo
meeting,
I
gave
a
quick
indication
of
of
I
wouldn't
say
problem,
but
a
certain
lack
of
smoothness
that
I'm
currently
experiencing
with
bringing
href
and
sibo
pact
together,
so
that
that's
also
something
that
we
probably
want
to
discuss.
D
A
A
And
and
most
of
the
remaining
feedback
required
from
implementation
was
about
the
authority
anomaly
if
I
remember
correctly
right
yeah,
so
we
really
need
to
close
that
point,
possibly
add
test
vectors
and
then
in
principle
the
document
would
be
finished.
D
I
think
so
it's
just
another
thing
on
the
to-do
list.
Sure
the
the
problem
with
this
young
zebra
stuff
is
really
that,
with
a
certain
amount
of
load,
one
switches
to
shortest
job
next
processing
and
charges
drop.
Next
processing
has
the
well-known
property
that
it
is
the
best
property,
the
best
processing
scheme
to
get
to
to
minimize
average
waiting
time,
but
also
it
means
you
get
starvation.
D
Yeah,
as
I
said,
I
I
think
we
we
have
it,
but
we
have
to
validate
that.
A
A
A
Yeah
moving
to
coral,
I
can
start
with
an
update,
because
christian,
crystal
and
thomas
could
not
be
here
today,
unfortunately,
but
the
work
is
continuing.
There
are
also
design
team
meetings
for
that,
mostly
by
the
authors,
in
fact,
and
there's
a
lot
left
to
do
here
on
course,
but
it
was
a
decision
to
focus
especially
on
two
main
points
in
the
short
mid
term.
A
A
We
actually
plan
to
have
additional
design
team
meetings
more
specific
on
on
this
point,
to
try
to
align
the
coral
information
model
to
to
sdf,
so
it'd
be
good
to
have
some
some
separate
design
team
meetings
on
this
aspect
involving
michael
koster
and
other
people
interested,
especially
in
that,
so
that
will
come
soon
and
the
second
high
priority
item
to
cover
is
the
diagnostic
notation
and
there
are
some
building
blocks
more
or
less
under
ongoing
work
can
help
and
in
fact,
karsten
mentioned
something
along
that
direction.
A
They'll
work
greatly:
can
you
just
easily
share
your
screen.
D
A
D
B
C
D
D
Yeah,
so
we
have
the
the
representation
that
that
we
really
care
about,
because
that's
what
what
our
applications
will
will
actually
work
on
and
we
have
a
bespoke
text-based
representation
that
is
great
for
whiteboard
and
rfcs,
but
it
has
all
those
problems
that
we
know
from
handcrafted
examples
that
then
you
suddenly
start
to
optimize
for
for
the
text-based
representation,
that's
just
how
how
our
brains
work
and
that
that
is
makes
it
too
easy
to
to
lose
sight
of
what
we
are
actually
trying
to
do,
which
is
to
get
the
interchange
get
the
the
working
with
the
interchange
representation
easy.
D
D
So
the
the
overall
syntax
is
patterned
after
the
the
existing
diagnostic
notation
syntax
for
writing
up
byte
strings,
where
we
expect
the
the
diagnostic
notation
tool
to
transform
between
the
actual
byte
strings
and
hex
of
a64
encoding,
and
here
we
would
expect
the
tool
to
actually
transform
between
cris
and
and
the
human
readable
ui
notation
for
that.
D
D
Of
course,
this
also
means
that
the
implementations
of
diagnostic
notation
may
have
to
be
mixed
in
with
the
code
for
for
href,
but
yeah
that
that's
the
the
problem
of
having
an
application,
specific
extension
by
the
way,
if
that
syntax
looks
familiar
to
you
from
from
coral.
D
Jaws
actually
proposed
doing
something
similar
for
daytime
objects.
So
we
may
just
keep
that
part
of
the
bespoke
text
notation
and
and
put
it
unchanged
into
the
application.
Specific
extensions
and,
of
course,
being
able
to
notate
a
daytime
is
something
that
is
not
really
very
application
specific
at
all,
so
that
this
might
become
a
rather
common
use
of
diagnostic
notation.
D
D
We
might
still
want
to
have
a
bespoke
text
notation
at
any
point
in
time
that
can
be
added
at
any
time.
So
we
are
not
completely
giving
up
on
the
idea,
but
we
are
giving
up
on
the
the
idea
of
having
it
right
out
of
the
box
and
we
make
sure
that
the
the
sibo
diagnostic
notation
can
actually
represent
coral
documents
nicely,
and
the
next
step,
of
course,
is
to
make
sure
that
the
sebor
pact
approach
actually
works
well
with
the
the
curry
mechanism.
D
That
has
been
part
of
coral,
so
that's
actually
the
the
rest
of
the
slides
or
the
slide
seven.
D
D
If
you
have
something
like
like
scheme
authority
player,
called
s,
slash,
slash,
co-op
dot
me
is
this:
game
is
an
authority.
So
these
are
the
two
first
elements
of
the
href
array
and
you
can
just
use
existing
zero,
packed
mechanisms
to
append
a
path
array
to
that
and
you
have
a
valid
ci,
but
it's
much
harder
to
do
this
for
the
next
prefix,
which
looks
very
similar
in
the
curry
notation.
D
But
if
you
look
at
this
from
a
ci
point
of
view,
it's
essentially
a
three-element
array
with
a
scheme,
an
authority
and
a
path,
and
we
don't
have
a
way
in
the
packed
mechanism
to
actually
reach
into
that
array
and
append
something
to
the
the
inner
array.
The
embedded
array
that
already
has
a
foo
so
to
get
slash,
fuselage
bar
into
a
ci.
D
You
would
need
to
to
say
go
into
that
array
and
and
add
bar
to
that.
So
that's
something
where,
where
I
don't
know
how
we
will
get
that
element
of
of
curries
in
there.
D
The
same
is,
of
course,
the
true
if
there
is
also
fragment
syntax
and
so
on
so
some
cases
of
of
doing
that
work
easily,
for
instance,
adding
the
actual
fragment
identifier
that
works
easily,
but
adding,
for
instance,
a
second
query
parameter
that
again
doesn't
work.
A
D
So
the
the
there
are
two
ways
to
to
handle
this
one
is
to
handle
it
outside
in
by
by
essentially
saying
this.
This
is
a
pact
reference
and-
and
we
we
reach
into
that
data
structure
to
put
stuff
in
there
or
it
can
be
handled
inside
out
by
identifying
in
the
data
structure
where
it
actually
is
meant
to
be
extended.
It's
a
little
bit
like
like
christian
circumfix
compression
that
he
proposed
for
sibo
pact.
D
So
that's
one
way
of
addressing
it.
It's
not
going
to
look
as
pretty
as
curry's,
but
curries
have
their
own
little
problem
because
they
they
are
essentially
doing
string,
concatenation
and
then
running
the
parsing
on
the
concatenated
string.
They
really
they
don't
have
actual
semantics.
So
you
only
know
the
semantics
after
having
passed
the
strings,
and
we
could
have
a
more
semantic
way
of
of
doing
curry
like
thing
with
the
eyes.
A
Okay
and
then
we
can
move
to
the
next
item.
It's
the
new
proposed
charter
text,
so
it
was
promised
to
prepare
a
first
proposal
at
itf
111,
the
just
prepared
it
and
it's
circulated
on
the
list
and
on
a
cody
md
note
both
linked
here
in
these
minutes.
A
As
far
as
I
know,
I
couldn't
see
any
feedback
or
comments
yet,
so
we
definitely
need
feedback
and
comments
from
the
group
before
proceeding
with
this
in
any
sense
but
like
intended
at
the
last
itf
meeting,
it's
not
a
revolution,
it
is
the
same
overall
direction
and
it
was
more
about
updating
the
text
to
reflect
what
has
actually
happened
and
and
what
is
happening,
and
most
of
that
was
about
touching
the
second
half
of
the
current
charter.
A
Actually,
the
first
part
was
mostly
an
editorial
revision,
so
we
really
need
input
from
the
group,
even
if
you
think
unlikely
that
everything
is
fine,
as
is
please
say
so,
on
the
list
or
commenting
in
line
in
the
codemd,
where
you
need
to
log
in
to
be
logged
in
to
comment,
and
we
plan
to
bring
this
up.
Also,
the
next
entry
meetings
to
check
the
progress
of
this,
and,
of
course,
if
you
have
early
comment
input
already
to
share
today,
please
do.
D
Yeah,
so
I
I
didn't
really
do
a
gif
of
the
two
yet
so
whatever
comment
I
have
is
is
more
on
the
meta
side,
so
when
co-op
was
actually
approved,
we
had
very
unlikely
number
of
yes
votes.
I
think
it
was
eight
yes
votes
or
something
I
don't
remember
the
actual
number,
but
something
that
hadn't
happened
before
so
we
used
to
have
an
isg
that
was
very,
very
supportive
of
doing
the
co-op
work
and
yeah
right
now.
D
I
think
we
have
an
isg
with
several
people
that
that
are
well
kind
of
annoyed
of
of
the
fact
that
we
sometimes
get
to
do
things
that
that
the
big
web
doesn't
get
to
do,
and
so
they
they
they
are
loaded
down
by
their
baggage
and
we
just
go
ahead
and
and
sprint
around
them
nimbly.
D
So
I
think
we
we
have
to
be
a
bit
more
careful
what
we
do
there
so
not
not
to
lose
even
more
support
in
in
the
isg.
D
So
that
that's
my
my
general
observation
and
that's,
why
I
said:
do
we
really
have
to
do
a
recharter
at
this
point
in
time?
But
maybe
you
can
remind
us
why
we
want
to
do
this.
A
It
started
out
of
spontaneous
considerations
among
chairs
and
the
id,
and
the
two
main
reasons
also
stated
at
the
last
itf
meeting
was
yeah
to
make
things
easier
and
safer
with
the
isg
actually
because
it,
it
became
more
often
the
case
that
they
could
bring.
Argument
like
this
is
not
in
the
charter
or
is
not
well
aligned
with
the
charter
yeah,
and
it
would
also
be
helpful
for
newcomers
to
have
an
up-to-date
idea
of
what
is
going
on.
B
Some
of
the
work
is
kind
of
at
the
limit
of
the
charter,
and
and
that
can
create
delays,
because
then
I
have
to
take
this
on
with
the
isg
and
convince
everybody
that.
F
B
It's
at
the
limit,
but
it's
still
in
charter-
and
I
don't
think
anything
we're
doing
is
not
in
charter,
but
we
could
remove
that
type
of
objections
to
the
work
we're
doing
if
the
charter
is
reflecting,
what
we're
doing
better.
D
B
Then
again,
I
I
don't
have
experience
with
the
with
the
rechartering
process
and
in
the
sense
that
I
don't
know
how,
how
opposed
to
to
rechartering
the
certain
isg
would
be.
So,
of
course,
we
don't
want
to
like
add
work,
and
we
don't
want
to
have.
B
D
A
A
Okay
and
we'll
solicit
more
also
for
for
additional
feedback
on
the
list
again.
A
Well,
some
documents
were
mentioned
still
as
draft,
including
block
wise
or
score
wasn't
even
mentioned.
A
There
was
a
vague
mentioning
of
using
cozy
or
investigate
the
use
of
cozy
like
things
that
happened
for
josie
before
I
think
t2
trg
was
to
mention
as
a
proposed
research
group
and
href,
and
coral
well
weren't
mentioned,
and
I
couldn't
see
something
that
that
was
in
a
safe
way,
letting
imaging
about
href
and
coral.
A
D
Good,
okay,
so
that
needs
to
be
checked
offline
and
just
because
I
didn't
have
a
good
way
of
dipping
it.
I
didn't
do
that
yet.
C
I'm
thinking
if
we
could
backtrack
from
the
google
doc
to
the
original
document
and
then
share
that
as
well.
Maybe
that's
easier
to
check
the
changes.
B
Would
be
a
good
idea?
Add
one
repository.
B
C
We
could
we
could
upload
the
current
charter
to
github
and
then
just
copy
paste
one
by
one.
All
the
previous
changes
we
made,
including
the
editorial
ones
so.
F
Yeah
I
found
that
the
group
security
work
was
well,
it's
very
important,
but
it
was
also
very
intricate
and
I
don't
know
for
done
the
step,
the
stage
of
it,
but
I
I
think
that
I
think
that
at
the
very
least
it
it
should
maybe
get
a
more
clear
design
team
to
iterate
on
things.
F
And
present
the
results
rather
than
try
to
design
them
in
the
within
the
working
group
constraints
of
time.
F
Doesn't
have
a
clear,
it
doesn't
have
a
clear
of
clear
focus
because
it's
done
its
core
work
and
it
has
a
whole
bunch
of
little
things
and
probably
yeah.
It
probably
just
needs
to
to
to
finish
its
work
and
shut
down,
not
try
to
adopt
something
new,
but
I
guess
I
would
say
that
the
kind
of
like
I
want
to
point
out,
I'm
sorry.
I
was
had
another
meeting
that
took
up
the
previous
45
minutes.
F
D
F
Yeah
so
I
agree
with
you
and
then
we
get
into
scheduling
conflicts
and
we
don't
get
enough
people
because
there's
too
many
conflicts,
and
so
that's
why
I'm
saying
maybe
that's
why
I
recanted
as
I
wrote
it
and
thought.
Well,
maybe
it
just
needs
to
have
a
more
structured
design
team
around
it
so
that
it
has
the
time
to
get
where
it
needs
to
go.
D
Yeah
the
network,
not
the
working
group,
has
a
design
team
on
versioning
and
labeling
that
has
been
meeting
I
think
weekly
for
a
year
or
so
so
that's
certainly
one
way
to
handle
complicated
problems.
D
So
the
the
main
reason
to
put
something
into
the
ace
working
group
is
to
employ
the
fact
that
it
has
a
security
id
behind
it.
D
F
That
I
think
that
that
they,
that
ace,
has
concerned
itself
a
little
bit
that
this
is
a
product.
My
impression
is,
this
is
a
product
that
is
designed,
for
instance,
for
sending
a
turn
on
all
the
lights
message
and
ace
has
certainly
got
the
authorization
behind
that
there,
and
I
I
don't
know
for
creating
I'm
unclear
whether
we
have
two
solutions
happening
or
complementary
ones
or
ones
that
build
optionally
on
each
other.
Euron
has
some
comments
too.
Here.
G
Yeah,
so
about
the
grupo
score,
there
is
actually
a
design
team,
so,
but
it's
not
like
presenting
itself
as
now
the
design
team
is,
is
presenting
it's
it's
results,
it's
more
like
there
is.
There
are
reviews
of
the
ongoing
work
by
a
set
of
people,
and
one
example
of
that
is
the
the
the
there
was
a
a
paper
or
sorry.
G
It
was
a
pre-print
by
by
eric
thor
mark
kerry,
from
erickson
on
on
the
properties
of
what
kind
of
properties
you
need
to
put
in
the
in
in
in
the
key
derivation.
What
kind
of
information
you
put
in
the
key
derivation
to
allow
certain
certain
attacks,
and
that
was
sort
of
led
to
a
change
in
in
group
score
during
the
spring?
G
So
there
is
actually
a
design
team
behind
this,
and
I
I
don't
see
the
benefit
of
moving
anything
of
that
work
to
ace
ac
is
doing,
as
you
said,
the
complementing
part,
which
is
how
you
get
the
keys
in
place,
but
it's
not
dealing
with
the
actual
communication,
security
or
security
for
this
for
groups
and
so
on.
G
D
No,
so
that
was
kind
of
off
topic
from
the
the
charter
discussion.
D
I
I
do
think
that
we
have
a
pretty
clear
delineation
between
ace
and
and
car,
so
getting
the
keys
in
place
is
maybe
not
exactly
the
way
I
would
put
it,
but
that's
at
least
a
good
first
approach
for
the
things
that
have
to
happen
in
ace
and
we
are
trying
to
make
sure
that
we
can
actually
communicate.
D
Based
on
that,
so
I
haven't
checked
the
the
charter,
whether
it
actually
reflects
this,
this
division
of
work,
so
that
is,
that
seems
to
be
a
pretty
obvious
addition
to
what
needs
to
go
into
an
updated
charter.
D
B
I
would
suggest
is
that
the
working
group
takes
a
look
at
the
list
of
documents
currently
adopted
by
the
working
group,
and
you
know
just
check
with
the
charter
and
see
how
well
this
fit,
because
that's
what
that's!
What
started
this
discussion
with
me
and
the
chairs,
and
I
could
foresee
some
some
discussion
in
the
isg
about
how
how
fitting
some
of
these
documents
are
with
the
current
charter
and
if,
if
the
working
group
believes
that
you
know
this
is
work
that
it's
not
worth
doing
a
rechartering
to
have
this
better
alignment.
B
I
it's
fine.
I
will
take
that
discussion
with
yesterday
when
it
comes
but
yeah
that
that's
that's.
I
think
what
needs
to
be
looked
at.
D
B
I
I
got
the
impression
that
this
isg
is
very
careful
with
with
the
checking
the
charter
and
it's
it's
more
about.
You
know
it's
it's
very
nitpicking
on
that
and
that's
why
I'm
bringing
this.
But
again
I
haven't
had
the
experience
of
how
painful
it
is
to
go
through
the
recharger.
B
D
A
Yaran
has
more
comments
in
the
chat.
Do
you
want
to
relay
them
you're
on
the
mic.
G
Well,
yeah,
I
just
posted
what
I
think.
I
think
it
was
the
latest
version
of
the
ace
charter.
So
I
think
there
is
a
nice
tie
in
there
to
what
what's
happening
here.
So
it's
talking
about
what
is
needed
for
applicability
of
group
communication,
it
doesn't
say,
put
the
keys
in
place.
It's
that
applicability,
which
is
better,
of
course,
it's
authorization
and
other
stuff
as
well,
and
then
it
says
also
that
it
applies
specifically
to
to
to
the
security
protocols
that
are
relevant
for
co-op.
C
A
F
C
Oh
no,
that
was
all
I
I
well
okay,
then
I
will
do
an
attempt
and
I'll
share
the
link
in
the
chat.
I
think
if
marco
is
okay,
we
can
do
it
on
this
repository
yeah
I'll
put
it
in
the
chat.
A
D
B
B
A
C
D
I
would
probably
take
the
the
original
charter
make
big
sausage
paragraphs
out
of
the
paragraphs
and
then
put
in
the
new
charter
again
in
the
form
of
sausage
paragraphs.
D
A
Right
so
this
new
draft
was
submitted
before
the
itf.
I
should
share
my
screen
or
my
camera.
A
This
draft
was
submitted
before
the
previous
itf
cut
off
following
a
number
of
discussions
started
in
other
documents,
and
it
basically
describes
how
to
use
all
score
at
proxies.
So
thinking
of
intermediaries
as
consumers
of
those
score,
and
why
is
that?
A
Well,
you
may
need
to
have
a
security
association
between
a
client
and
a
proxy
to
put
it
simple
for
the
number
of
reasons,
and
we
have
some
use
cases
where
that
makes
sense
or
is
need
to
happen,
and
they
are
in
the
draft
and
a
few
in
the
coming
slides.
A
The
moment
you
have
a
security
association
between
a
client
and
the
proxy
you'd
be
good
to
have
it
based
on
on
oscor2,
and
especially
in
the
case
where
you're
using
oscor
already
into
end
between
client
and
server.
So
for
the
sake
of
stack
simplicity,
code,
reuse,
keep
provisioning
and
so
on.
It's
just
better
to
to
use
all
score
again.
Also
in
that
communication
leg.
A
Now,
if
you
look
at
the
current
oscar
rfc,
of
course,
the
only
intended
consumers
of
all
score
or
oscar
endpoints,
as
we
call
them,
are
just
the
origin,
client
and
server,
of
course,
and
and
you
can
also
imagine
that
if
you
start
from
end
to
end
oscar
and
you
add
an
additional
score
protection
between
say
the
client
and
the
proxy,
you
end
up
protecting
a
message
twice
in
a
row
and
that's
not
only
undefined,
but
it's
actually
forbidden
today.
So
this
is
not
really
possible
right.
A
Now,
a
concrete
reason
to
do
that
started
in
another
document
group
comproxy,
where
we
started
to
think
along
these
lines
in
an
appendix
this
came
up
again
at
itf
110
and
at
the
following
interim,
and
there
was
some
agreement
about
having
this
particular
point
taken
out
from
that
document
and
we
better
explained
elaborated
scrutinized
in
a
totally
separate
draft,
also
because
it
seems
to
have
more
generic
applicability
that
in
that
group
comproxy
context
and
the
result
is
in
fact
this
particular
draft
that
we
made.
A
Eric
happen,
use
cases
is
also
summarized
in
the
draft
one.
I
mentioned
already
the
group
com
proxy
case
that
started
this
out
of
comments
from
from
christian.
Actually-
and
here
you
have
a
client
that
wants
to
reach
a
group
of
servers
through
a
proxy
that
practically
sends
out
the
request
over
multicast.
A
You
may
have
group
of
score
end
to
end
between
the
client
and
the
servers,
and
here
the
proxy
has
to
identify
the
client
before
just
forwarding
over
multicast
request
and
here
again,
a
security
association
and
the
advantage
of
having
it
based
on
oscar,
as
I
mentioned
before.
A
A
second
use
case
is
kind
of
still
related
to
group
communication,
but
in
in
a
different
way,
you
can
have
a
server
sending
a
multicast
notifications
to
multiple
clients,
observing
the
same
resource
and
those
can
also
be
protected
with
group
score
end
to
end,
and
you
may
have
a
proxy
deployed,
and
in
that
particular
case,
an
additional
step
is
required
by
the
clients
to
provide
a
particular
reference
request
to
the
proxy
to
put
the
proxy
in
a
position
to
to
do
the
job,
and
we
are
already
discussing
in
the
security
considerations
of
that
document,
observe
multicast
notifications
that
it's
just
very,
very
good
idea
to
protect
that
particular
reference
request
on
the
leg
between
client
and
proxy.
A
So
you
have
the
same
thing
again
and
third,
one
we
mentioned
is
related
to
lightweight
m2m
in
the
1.2
version
of
the
standard
thinking
of
the
oscore
case,
so
oscar
used
in
in
the
registration
process
to
the
life
within
2m
server,
it
is
mentioned,
as
oscor-
can
also
be
used
by
the
life
within
twin
client
to
establish
a
communication
with
an
application
server
external
to
the
lifeline
film
domain,
using
the
live
with
informed
server
as
a
proxy.
A
So
here
it
comes
again
the
life
with
informed
server
that
has
to
recognize
the
life
into
a
client
as
it
does
as
usual
anyway,
but
here
especially
for
forwarding
traffic
out
of
the
life
with
influence
domain,
and
you
end
up
performing
a
double
protection
of
that
message.
A
Again,
we
have
some
discussion
with
christian
also
that
suggested
something
more
like
how
this
can
help
for
third-party
proxies
using
oscor
for
the
new
transport
indication
work
he
proposed,
and
he
also
suggested
to
think
about
other
kind
of
proxies,
acting
like
network
entry
points
like
big
firewall
and
so
on.
A
So,
given
this
context,
the
contribution
would
be
about
updating
your
score
rfc
to
make
this
possible
and
the
update
before
coming,
to
defining
the
actual
mechanics
of
how
this
works
is
about
enabling
oscar
to
be
used
also
at
a
proxy
and
intermediary.
Just
like
origin,
client
and
server
do
and
also
explicitly
admit
what
is
now
forbidden
so
having
what
is
called
nested
l
score
in
rfc
8613
and
in
the
current
version
of
the
draft.
A
We
actually
limited
this
to
exact
two
oscar
layers,
possibly
applied
to
the
same
message,
but
out
of
some
more
discussion
with
christian
and
joran.
It
is
actually
possible
to
do
and
interested
to
explore
to
not
put
this
hard
limit
because
it
is
possible,
first
of
all,
with
no
particular
additional
effort,
and
it
can
be
actually
useful
in
some
very
particular
cases
to
be
explored
more
and
the
focus
here
is
on
all
score
in
particular.
But
you
can
take
this
as
is,
and
it
works
just
as
well
with
group
score.
2.
A
And
during
the
design
phase
of
this,
we
found
pretty
useful
to
come
up
with
the
notation
like
this,
and
we
got
also
among
the
feedback
that
that
was
fine
during
the
design.
But
it's
just
better
to
get
rid
of
something
like
this.
That
can
be
too
heavy
to
read
and
process,
because
this
in
the
draft
evolves
into
a
pretty
fine
grain
description
of
the
message
processing,
and
we
already
have
in
mind
how
to
work
in
that
direction,
but
to
better
understand
at
least
this
first
example
focusing
on
colors.
A
If
you
think
of
the
most
simple
example,
one
client
proxy
one
server,
you
can
independently
use
all
score
between
client
and
proxy,
yes
or
no
between
proxima
server,
yes
or
no
or
end
to
end
between
client
and
server,
which
is
the
only
possibility
admitted
today,
and
that,
of
course,
results
in
eight
possible
configurations,
as
we
call
them
again,
mostly
at
the
same
time.
A
To
be
sure
we
weren't
missing
any
any
corner
case,
but
we'll
hopefully
get
rid
of
most
of
this
heavy
notation
in
the
next
version
and
without
going
into
any
detail
at
all.
The
idea
is
that
the
client
can
first
of
all
protect
a
request
and
to
end
with
oscar
as
usual,
but
then,
if
it
has
also
or
only
enos
core
context
shared
with
the
proxy
and
it
can
protect
the
request.
A
Additionally,
also
with
that
context,
this
can
result
in
a
double
protection
as
a
particular
case,
and
the
the
interesting
deviation
here
is
that
some
of
the
options
are
really
intended
for
the
proxy
and
it
makes
sense
to
protect
them
when,
when
performing
the
the
red
protection
here,
in
particular
the
the
the
proxy
related
co-op
options
and
you'd,
be
also
about
protecting
the
oscar
option
that
was
produced
from
the
first
perform
protection
they
will
be
encrypted
and
the
result
of
the
second
protection
will
be
a
new
oscar
option
that
will
be
sent
as
outer
option
on
the
wire.
A
You
reach
the
proxy
at
some
point
and
we
realized
but
didn't
explicitly,
write
in
the
draft
that
we
are
not
adding
and
we
don't
need
to
add
any
particular
new,
explicit
signaling
as
a
hint
about
what
to
do
what
we
already
have
around
the
options
included,
and
their
combination
is
already
enough
to
understand
what
the
proxy
in
this
particular
case
is
supposed
to
do,
and
at
the
end
of
the
day,
it
either
forwards
to
the
server
according
to
what
the
proxy
options
say,
or
it
delivered
to
its
application.
A
If
it
has
any
at
all
or
well,
it
decrypts
the
message
according
to
what
the
oscar
option
says
and
possibly
repeats
this
cycle
depending
on
the
result
of
that
decryption.
A
So,
for
the
server
point
of
view,
it's
mostly
about
being
fine
in
receiving
something
which
is
possibly
double
protected,
and
then,
when
it
comes
about
producing
a
response,
it
has
to
think
in
a
similar
way
that
the
client
thought-
and
it
is
possibly
a
double
protection
if
you
involve
both
oscar
end-to-end
and
with
the
proxy,
the
proxy
things
in
a
similar
way
for
forwarding
back
and
at
the
end
of
the
day,
the
client
has
to
be
fine
in
accepting
something
double
protected
so
to
strip
out
of
the
message
two
oscar
layers
rather
than
one.
A
A
We
we
learned
especially
that
the
use
cases
look
good
and
there
can
actually
be
a
few
more
and
the
overall
process
makes
sense,
but
it's
presented
in
a
far
too
complicated
way,
especially
considering
yet
flags
and
yet
configurations
I
showed
before
so
out
of
their
suggestion.
The
plan
is
to
do
a
major
revision
of
the
presentation
of
the
message
processing
to
make
it
a
more
general
high-level
algorithm.
A
A
This
seems
totally
possible
and
it
can
lead
to
something
pretty
promising,
but
it
is
not
for
the
immediate
next
update
to
to
explore
fully
in
any
way.
A
Yeah,
so
the
plan
is
to
work
on
a
version,
one
especially
adopting
the
comments
we
we
got
and
especially
on
the
presentation
of
the
message
processing
to
make
it
easier
to
read
and
to
extend
to
a
chain
of
proxy
and
already
to
mention
possible.
This
additional
usage
of
this
and
yeah
start
thinking
on
the
triple
quadruple
application
of
oscar
layers,
so
we're
pretty
happy
with
the
feedback
so
far
already,
and
we
have
a
lot
of
clear
work
to
do.
Of
course,
more
comments.
Input
on
this
are
very
very
welcome.
D
D
D
The
the
shorter
range
processing
actually
can
not
understand.
What's
going
on
in
the
logarithm
correct,
so.
E
F
So,
are
you
assuming
that
all
proxies
are,
I
want
to
say
known,
I
mean
to
say
our
clearly
discovered
or
on
the
all
route.
F
Sorry,
you
say
next
hop
my
neck,
my
first
proxy
you
mean,
or
my
final.
A
F
It
I
I
have
to
be
no,
I
the
entry
of
the
train
has
to
be
known
to
me.
It
can't
be
that
it
just
inserts
itself
in
the
middle
and
proves
that
it
exists.
A
It
and-
and
I
expect
anyway,
most
of
the
cases
are
not
about
the
chain-
are
about
one
proxy,
but
you
never
know
you
may
have
a
chain,
you
would
work,
then
the
thing
goes
hope.
I
hope.
G
Just
to
comment
on
carson's
statement:
no,
no
one
is
saying
anything
so
assume
they
agree.
I
just
I
mean
I
this
is.
There
is
no
news
here
from
from
my
point
of
view,
this
was
what
you
intended
to
present
at
itf
111.
Sorry,
I'm
I'm,
I'm
I'm
not
saying
the
right
thing
most
most
of
the
content.
G
A
F
C
A
B
B
B
C
B
B
I
followed
up
with
martin
d,
martin,
duke
and
murray,
and
they
were
happy
with
the
resolution.
So
the
only
two
left
in
these
comments
are
roman
and
ben
and
ben
has
replied
to
to
christian's
answers.
So
it's
back
with
the
authors
and
roman
has
not.
I
don't
think
he
has
had
time
to
look
at
the
reply.
So
that's
the
status.
C
D
C
B
Okay,
yeah,
I
don't
see
that
in
my
list
of
documents,
but
that's
probably
because
it
was
before
before
me.