►
From YouTube: IETF101-CORE-20180319-1330
Description
CORE meeting session at IETF101
2018/03/19 1330
https://datatracker.ietf.org/meeting/101/proceedings/
A
A
So
this
is
things
Jason,
Oscar's
animal
resource
directory
and
said,
and
tomorrow
we
will
have
cocoa,
because
there
will
be
a
transport
area
of
person
coming
in.
That's
why
this
is
tomorrow.
We
will
there,
you
and
I
have
managed
to
put
this
into
other
slides
yet
and
then
we
will
go
through
documents
that
are
a
little
bit
less
advanced
and
you
can
see
there's
a
lot
of
documents,
and
so
this
will
be
little
bit
of
a
gong
show.
Tomorrow
we
have
any
changes
additions
to
the
agenda.
We
need
to
do
okay.
A
So
let
me
quickly
talk
about
the
working
group
status.
We
have
one
RFC
published
since
the
last
ITF,
which
is
the
coil
over
tcp
TNS.
That
was
a
pretty
long
last
fire
we
had
with
this
draft,
but
we
finally
got
it
published
on
February,
15
and
and
one
one
of
the
things
that
we
actually
had
to
do
is
write
another
RFC
that
that
was
just
pitching
a
minor
hole
in
the
circuits
environment.
B
A
A
A
Okay,
so
this
I'm
not
going
through
the
milestones
right
now,
because
it's
three
documents
to
iteration,
so
we
will
talk
about
the
documents
when
we
talk
about
the
recognition
but
I
have
a
couple
of
advertisements
at
the
beginning.
First
of
all,
there
will
be
I
think
you
think
we
research
group
meeting
on
coexistence
this
afternoon.
Those
of
you
who
were
the
hot
FC
talks
have
Laura
talked
about
the
problem
she's
trying
to
address
so
right.
Now
we
are
building
cars
under
the
assumption.
A
There
are
no
other
cars
on
the
highway
and
that
works
for
a
while,
but
at
some
point
we
will
have
to
put
in
Direction
indicators
and
all
the
other
things
that
allows
cars
to
work
in
traffic,
and
so
coexistence
is
mostly
seen
as
a
fine
layer
thing,
but
that's
probably
not
true.
We
have
to
think
about
coexistence.
How
does
our
pet
network
work
in
an
apartment,
building
together
with
200
other
networks?
A
That's
a
problem
that
we
currently
don't
quite
know
how
to
solve
so
its
research,
but
I
think
it
might
be
interesting
to
do
this
as
well.
Second
thing:
there
is
a
draft
that
is
going
to
be
workgroup
last
called
very
very
soon
in
the
six
dish
working
group,
and
that
draft
contains
a
co-op
option
and
that's
I
think.
The
way
it
should
be
quote
is
designed
for
extensibility,
so
other
groups
should
be
able
to
define
options.
It's
an
option
that
we
have
talked
about
here.
A
The
stateless
proxy
option
and
I
would
advise
everyone
who
who
cares
about
things
like
this,
to
look
at
the
draft
and
maybe
send
a
message
to
call
and
or
six
dish
about
any
comments
that
you
have
and
maybe
go
to
the
meeting
on
Wednesday
and
Viscount,
who
so
just
to
give
away
what?
What
I'm
going
to
write
I
think
we
had
a
discussion
on
whether
that
option
could
be
repeatable
or
not
right.
A
A
So
please
go
there
if
you
care
about
resource
directory
and
or
DNS
system,
okay,
and
that
one
message
from
our
sponsors,
this
is
really
not
directly
related
to
to
call.
But
it's
interesting
to
everybody
who
cares
about
constraint
system,
so
the
next
right
savage
will
be
in
September
and
sediment
you,
you
might
care
about
that.
A
A
Thought
well,
it
might
be
a
good
idea
to
have
a
JC
to
avoid
having
the
need
for
another
pass
I
in
mid-2015
we
added
seaboard
variant.
So
the
whole
point
is
about
documents
that
contained
links,
and
it
is
used
for
self
description
of
car
notes.
It's
in
flesh
we're
now,
instead
called
and
for
a
long
time
we
have
been
working
on
this
document,
with
the
assumption
that
the
focus
should
be
on
round
trip
abilities
with
RFC
66.
A
Ninety,
and
that's
actually,
the
that
being
the
focus
of
the
draft
that
passed
burden
of
last
fall
last
year
and
was
submitted
to
the
iesg
on
April,
2nd
and
then
gates,
and
we
got
lots
of
feedback
and
there's
also
related
work,
for
instance
in
BMC
F
spec
that
we
probably
shouldn't
ignore.
So
they
are
not
using.
Is
they
same
forward,
but
it's
kind
of
culturally
compatible
and
we
should
try
to
understand
what
they
were
trying
to
do.
A
So
the
the
idea
is
to
have
a
full
web
linking
environment
in
the
JSON
or
support
variants,
but,
of
course,
still
cover
all
of
RFC.
Six
is
ignited
because
we
at
least
want
to
make
sure
that
the
forward
direction
works,
even
if
the
backward
direction
may
not
always
be
possible,
also
communications
there.
A
So
that
is
the
the
we
focus.
If
you
look
at
the
document,
there's
actually
not
much
change.
I
have
submitted
event
that
does
most
of
the
changes
already,
so
it's
still
pretty
much
the
same
document.
There
are
some
details
that
are
now
refocused
on
this
and
there
is
one
background
activity
that
happened.
While
we
were
last
calling
the
JSON
document,
an
update
of
RFC
59
88,
which
is
the
base
level.
Obviously
66
90
happened.
A
That's
probably
not
a
big
problem
because
they
were
careful
of
not
trying
to
break
too
many
things.
But
again
what
happens
is
they
have
to
refocus
on
it
little
bit
so,
for
instance,
the
the
way
a
B
and
F
was
used
in
59
88,
which
creates
some
of
the
little
problems
we
are
having
here
that
was
fixed
in
82
88.
A
A
So
there
is
a
few
places
where
this
comes
up
and
I
just
put
in
a
one
here
to
the
slides
language
tags.
Our
c59
8h
defines
some
some
special
target.
Attributes
that
have
a
star
an
asterisk
in
their
name
and
what's
special
about
them
is
not
actually
the
attribute
itself,
but
the
way
they
are
represented
in
an
HTTP
link
header
because
it
should
be
linked
at
us-
may
need
to
be
a
ski.
A
So
you
know
whether
some
some
Chinese
characters
are
in
reality,
Japanese
practice,
so
66
90
has
some
support
in
its
a
B
and
F
for
a
title
star
attribute,
but
that
uses
exactly
the
weird
HTTP
link
kind
of
syntax,
and
originally
we
just
copied
that
over
into
links
Jason
and
it
just
doesn't
make
sense-
you
shut
up
cause
decoder.
Is
this
stuff
and
put
it
in
an
extra
Unicode
and
the
new
thing
is.
It
became
clear
in
the
discussion
that
we
probably
also
need
to
preserve
the
language.
A
A
A
title
then,
the
equal
spring
and
put
in
your
language
take
it
helps
to
okay.
So
this
is
the
new
stuff
and
we
all
have
to
look
at
it
and
we.
This
is
the
right
way
to
do
it.
Well,
that's
that's
a
detailer,
but
on
a
high
level
is
this
right
form
right
way
forward?
We
they
see
the
document
on
other
places
where
each
wait.
It
has
cleaned
up.
Things
explain
how
become
links
placed
in
touch
so
make
sure
the
nothing
is
fully
defined
in
that
direction,
but
otherwise.
A
690
using
crystals,
that's
one
and
another
direction
that
we
as
very
know
should
be
deciding
at
some
point.
Is
there
and
intention
of
getting
rid
of
in
college,
who
actually
want
to
replace
when
you're
caught
by
a
Jason
for
more
Zeebo
form?
That's
a
second
thing
which
does
not
have
a
bearing
on
what
we
do
in
the
customer,
but
there's
something
that.
A
B
So
is
actually
armed.
First,
three
bullet
points.
I
think
this
is
completely
logical
as
a
way
forward,
so
I
don't
have
any
pushback
there
I
not
intimately
familiar
with
the
ice-t
comments
around
those
if
this
fulfills
those
requirements.
So
it
sounds
like
Carson's
point
of
view
it
does
the
last
one.
You
know
I
would
comment
that
we
have.
B
We
have
media
type
variations
with
different
civilizations
for
a
reason,
I,
don't
see
any
reason
why
these
can't
coexist,
but
this
needs
to
be
made
clear
right,
like
if
I
request,
a
see,
bore
version
of
thought
well
known,
slash
core
I
should
get
that
rather
than
an
implementation
just
blowing
up,
because
what
you
should
get
RC
six
to
six
ninety,
why
are
you
asking
Rivers
for
Seaguar,
alright,
so
I
think
we
need
to
be
able
to
be
future
proof
also
with
how
we
deal
was
slash
download
on
slash
course.
B
A
B
Don't
think
we
have
I,
don't
think
we
there's
a
difference
between
mandatory
to
implement
recommended
right,
I
mean
you
need
something
that
everyone
has
to
support
right
now.
You
know
the
wave
flash
sizes
and
and
meta
devices
are
increasing
at
that
rate.
C
bar
is
not
an
issue
right.
We
have
more
issues
with
overhead
on
the
wire
energy,
etc.
So
I
don't
see
C
bore
as
a
limitation
in
these
devices,
but
we
do
have
enough
stuff
out
there.
That's
using
66
90,
you
know
it's
in
the
wild.
D
Yeah
Michael
Koster,
smart
things
in
Rd,
I
think
the
approach
we
took
and
I
Christian
can
correct
me.
If
it's
that's
been
changed
was
that
we
required
RFC,
66,
90
compatibility
and
allowed
additional
media
types
and,
if
that's
what's
being
proposed
here,
I
think
that
makes
sense
for
well-known
core.
Also.
A
A
E
A
A
F
G
Hello,
everyone
I'm
Yolanda
from
erikson
I'm,
going
to
present
the
next
update
on
our
score,
so
we
are
now
in
version
11.
My
co-authors
I'd
like
to
go
through
the
issues
from
from
the
IC
review.
Here
is
the
status
number
of
implementations.
We
actually
appeared
one
new
implementation
at
the
end
of
the
list
there,
which
so
it's
really
being
used
several
in
crops
up
in
time.
But
topic
of
today's
presentation
is
about
the
ISD
your
view.
G
G
So
there
are
a
few
slides
on
on
on
review
comments
I'm
not
going
through
all,
but
those
that
I
think
are
representative
port
for
the
type
of
comments
we've
got
and
here
is
a
comment
saying.
The
document
needs
to
security
analysis,
section
and
implications
and
modifications
of
unconnected
fields
are
are
not
sufficiently
covered,
so
they
written
was
already
a
security
consideration,
but
the
IC
wanted
to
have
more
specifically
an
analysis
of
the
security
that
is
being
used
in
the
protocol
and
what
guarantees
are
provided.
G
So
what
we
propose
to
do
is
add,
and
what
we
already
have
done
is
to
add
an
appendix
describing
security
properties
of
the
protocol.
There
are
two
three
sub
sections.
One
is
looking
at
the
problem
statement,
the
assumptions
from
even
videos,
so
that
was
not
really
clear
to
everyone
making
the
review.
So
we
think
that
needs
to
be
highlighted.
G
That
means
the
media
are
supposed
to
perform
certain
actions
using
the
available
headers,
but
other
other
message.
Fields
are
not
available
to
the
intermediaries,
so
there
are
protective
headers
that
are
unpredictable,
and
then
we
will
also
include
two
sub
as
one
which
is
going
through
the
security
properties
of
the
protected
headers
and
one
which
is
going
through
the
unprotected
fields
and
the
consequences
of
being
operated,
and
there
were
questions
around
the
nonce
construction.
G
Why
is
Sandra
include
nonce
and
the
rationale?
A
is
that
this,
the
security
design,
is
you
it's
intended
to
support
both
the
ordinary
request
response
notifications,
as
well
as
an
interchange
of
client
and
server
roles,
and
that
required
a
special
construction.
So
what
we
think
is
needed
for
for
understanding
this
better
is
that
we
have
a
proof
for
how
quino's
uniqueness
is
obtained,
and
we
put
that
as
fourth
subsection
in
the
new
context.
G
A
We
can
quicken
on
it,
but
I
think
it
would
be
useful
to
start
a
document
that
just
explains
different
situations
in
which
Oscars
being
used.
So
people
understand
why
that
is
actually
not
not
true.
I
mean
it
should
not
go
into
this
document
or
to
finish
it,
but
having
a
short
document
that
even
might
go
away.
I,
don't
know
that
just
says
these
are
different
configurations
in
which
we
run
or
score,
and
these
are
the
rationales
for
why
we
want
to
support
these
configurations.
I
think
that
would
be
used
for
Lynch
good.
A
G
G
A
A
A
A
G
G
The
host
comment
was
saying:
this
is
a
partial
solution,
because
there
is
no
key
exchange
protocol,
but
there
are
other
ways
to
establish
keys
besides
key
exchange,
and
so
you
could
either
do
it
with
with
the
the
probe,
the
Oh
score
profile
of
the
ACE
framework,
or
you
saw
my
deployments-
require
PSK
so
because
of
their
constrained
nature.
So
it
is
a
complete
solution
but,
on
the
other
hand,
the
key
exchange
protocol.
It
is
also
something
that's
needed
for
Osborne
that
was
discussed
in
the
ACE
working
group
this
morning.
G
So
so
that's
so
we
we
agree
with
that.
You
need
for
a
key
exchange,
but
it's
not
a
necessary
condition
for
all
school
and
then
there
were
a
bunch
of
comments
of
HTTP
and
they
were
very
good
because
we
hadn't
focused
so
much
so
first
comment
this
protocol
abuses
HTTP
by
tunneling
over
it
and
that
it's
actually
correct,
and
that
was
exactly
what
was
intended.
G
We
had
a
missing
a
be
an
effort
and
that's
fixed
now
yo,
and
it
was
a
question
about
how
how
does
the
coop
HTTP
gateway
understand
the
significance
of
the
new
header
field.
So
that's
the
obvious
security
header
field
and
insert
the
medium
pipe.
When
translating
and
that's
exactly
what
we
wanted
to
intend.
So
we
don't
want
to
necessarily
add
at
the
content
format
for
knows
a
bit
corny
obviate
security,
protective
message
so.
A
But
custom
from
the
floor
I'm
not
sure
we
are
actually
abusing
HTTP.
Yes,
we
are
tunneling
through
it,
but
we
are
using
it
as
a
request
response
protocol.
It's
not
like
we
are
just
using
HTTP,
do
to
send
messages
back
and
forth,
so
I
think
there
are.
There
are
different
levels
of
it,
abused
and
I
think
we
are
not
in
the
murder
range
here.
We
are
just
in
the
manslaughter.
G
Thanks
for
the
clarification,
so
another
Winston
question
was:
do
we
need
a
code
point
for
coop
for
this
new
media
Titan?
And
we
had
this
question
before
and
we
said,
as
we
said
that
time
previously,
we
said:
no,
we
actually
don't
need
it
for
this
draft
and
then
we
realized
that
maybe
someone
will
actually
have
a
use
for
for
this
code
form.
So
we
decided
to
include
that
anyway,
but
not
not
use
it
in
the
draftin
and
explain
when
you're
not
going
to
use
it.
So
that's
a
placeholder
currently.
G
What
if
the
request
is
redirected
by
the
server
that
doesn't
understand
those
core?
Ok,
so
here
are
two.
These
two
questions
are
actually
still
things
which
we
would
like
to
have
good
feedback
from
the
working
group.
Shall
we
support
HTTP
redirects
and
the
setting
is
two.
There
are
two
settings
one
setting.
This
is
one
of
the
settings.
Http
redirects,
you
have
a
coop
through
an
HTTP
proxy
coop,
HTTP
proxy
and
you
get
counter
server
and
the
server
makes
the
redirect
back
to
the
proxy.
That
would
be
the
ACD.
No,
it's
not!
Okay,
let's
not
recognize.
G
H
G
Good
question
I
think
the
reason
why
I
mean
this
is
one
of
the
questions
we
got
from
from
the
ISC
review,
so
we
assumed
that
this
was
reasonable
to
address
in
the
doc,
but
the
answer
could
be
that
it's
not
something
that
we
are
actually
in
this.
There
is
also
by
the
way
when
you
define
a
new
HTTP
header.
You
need
to
answer
a
number
of
questions.
How
will
will
read
is
be
impacted
by
redirect
so
I
think
it.
G
I
G
For
so
we
have
a
media-type
before
for
for
over,
for
the
object
security
for
the
payload
containing
the
compressed
cozy
object.
We
have
a
media-type
for
that
and
the
question
is:
do
we
also
need
a
content
format,
a
value,
and
we
don't
need
it
for
this
draft,
but
maybe
someone
wants
to
use
this
compressed
cozy
object
and
pass
along
a
media
type
for
it.
Okay,
I
don't
have
a
better
I,
better
answer
than
that
people
have
been
requesting
this
and
we
thought
it
better
include
it
now.
A
So
maybe
to
answer
this
question
the
there
is
no
need
to
include
the
content
for
adoption,
because
the
object
security
option
already
implies
the
value
of
the
content
for
adoption.
However,
if
this
media
type
is
used
in
some
composite
media
type,
that
has
content
from
it
numbers
on
the
specific
elements
that
go
into
the
composite
media
type,
then
it
is
useful
to
have
a
number
for
that.
That's
all
that's
the
whole
reason.
G
Next
ask
him
sort
of
a
while,
while
thinking
about
that
so
yeah,
so
the
question
actually
was
then
brought
got
brought
up
again
by
Dave
Taylor.
So
he
said.
Has
the
working
group
made
a
conscious
decision
here,
whether
we
going
to
support
HTTP
redirects
or
not?
He
was
not.
He
didn't
really
care
about
the
answer,
but
he
just
wanted
the
the
working
group
to
have
take
the
decision
that
either
no
we
are
not
supporting
a
stirry
redirects
or
yes,
we
will
support
them
and
we
will
make
sure
that
the
necessary,
whatever
specification,
is
in
place.
F
Aleksey
I
will
try
to
provide
nori,
not
very
formed
and
not
necessarily
correct
opinion
on
this
I
was
talking
to
Patrick
McManus
one
over
HTTP
base
coach
Ayers
about
another
document
that
decided
to
prohibit
redirects,
and
he
said
basically,
you
shouldn't
be
doing
this
when
you
are
using
HTTP.
The
error
direct
is
a
standard
feature.
You
should
just
support
it.
Basically,.
F
F
Don't
think
that
actually
I
need
to
think
a
bit
more
about
this,
but
I
think
you
can
have
HTTP
proxies
on
the
path
that
don't
have
to
understand
or
score
as
long
as
they
in
implement
HTTP
correctly,
they
will
work
with
what
you
know:
it's
just
another
payload
for
for
for
them,
so
they
can
redirect
another
source
and
it
can
go
to
another
server
that
is
HTTP,
2,
core
proxy
and
and
then
everything
should
be
fine.
So
my
gut
feeling
I
don't
think
there
is
anything
inherently
wrong
with
using
redirects.
J
Franchesca,
so
the
problem
we
are
thinking
of
is
that
if
it
doesn't
reach
the
right
server,
the
server
won't
like
the
security
material
won't
be
there.
If
it's
if
it
reaches
a
different
server,
it's
not
a
proxy
problem,
it's
a
fluid.
Who
is
this
intended
to
and
do
they
have
the
right
security,
material,
I.
I
G
So
we
will
make
sure
that
this
support
HTTP
redirects
in
some
way,
and
you
don't
need
to
specify
something
we'll
find
out
what
you
what
it
appeared.
The
last
point
on
this
slide
is
the
name
of
the
HTTP
header
field.
So
originally
we
call
it
object
security
and
we
got
some
comments
that
we
should
probably
be
more
specific
that
it's
not
it's
actually
dealing
with
constrain
the
restful
setting.
So
the
proposal
was
to
change
to
coop
object
security,
which
is
the
current
text
in
the
drafts,
and
now
we
got
another
proposal
from
Dave
Taylor.
G
I
C
G
A
The
option
name
never
occurs
on
the
wire,
so
you
don't
get
in
a
situation
where
you
need
to
google
it,
but
the
the
HTTP
header
mean
that
occurs
on
the
wire,
and
if
you
find
a
strange
baguette
where
that
in
it,
you
might
want
to
move
it.
So
I
think
that
the
Google
ability
of
the
header
field
name
is
a
is
an
important
thing:
how
we
actually
call
the
the
option.
We
have
very
wide.
G
Then
we
move
on
to
that
was
the
last
actually.
So
this
is
not
a
complete
cover.
Ii
of
all
other
your
comments,
you
could
look
at
the
wiki
and
the
commits
that
we
have
yeah
I'm
gonna
get
off
essentially,
but
in
summary,
we
have
tried
to
clarify
the
points
which
were
brought
up
and
fix
all
the
editorials,
and
there
is
a
new
appendix
or
reviewing
which
is
going.
Has
these
four
subsections,
as
we
discussed
yeah,
but
essentially
with
version
11.
G
J
Francesca,
so
if
you
take
a
look
at
the
wiki
in
the
github,
we're
really
tracking
all
the
issues
and
what
we
have
done
and
we
consider
we
just
have
to
answer
and
what
we
need
to
just
review
in
general
I
think
we
all
the
all
the
elements
are
in
the
draft.
We
might
want
to
rephrase
some
things,
but
the
content
is
there
in
our
opinion.
And
yes,
so
we
just
need
to
answer
the
emails.
We
got.
G
K
K
Okay
stages,
first
of
all,
we're
done
from
burning
crow
point
of
view,
so
sentimental
is
currently
I
get
last
cold
yeah.
That
job
is
scheduled
one
month
from
now,
so
there
was
a
since
last
person.
Alex
I
did
a
thorough
review
on
it
and
some
bunch
of
editorial
or
clarifications,
etc
that
we
fixed
there.
One
that
you
could
consider
a
technical
change
was
the
media
type
registration
for
exi
was
changed
from
plus
EXI
to
minus
CX
I,
although
yesterday
we
did
discover
this
may
be
worth
clarifying.
K
K
Then,
for
early
assignments,
so
this
currently
the
verse
that
is
submitted
has
a
bunch
of
DVDs
on
the
IDS
course
that's
gonna
be
assigned
by
iana,
but
since
there's
already
a
lot
of
use
for
cinema
will
be
useful
to
have
those
IDs
assigned
as
soon
as
possible.
So
what
we've
been
discussing
now
with
reviewers
expert
reviewers
for
that
registry,
is
that
okay,
it
is
a
good
set
of
ID's,
have
a
one
consequent
set
of
ID's
for
all
that
all
the
numbers,
except
for
XML.
K
So
the
reason
for
those
XML
IDs
being
in
the
two
pide
range
is
that
XLS
anyway,
quite
verbose,
you're,
not
saving
much
or
have
using
the
one
byte
IDs,
so
so
length
read
with
the
experts
on
this
ante.
We
all
seem
to
be
having
consensus
on
this,
so
we'll
be
going
forward
with
these
assignments.
Unless
someone
sees
any
concerns
so.
F
I'm,
trying
to
remember
I
might
not
have
added
it
to
the
next
to
each
other.
It's
probably
going
to
be
a
nice
chair
review
in
April.
K
F
K
K
F
K
Okay
thanks,
then,
there's
another
question
on
early
assignments:
how
about
extensions
to
send
a
mail?
So
since
it's
not
an
RFC,
yet
we
don't
have
the
registry
actually,
but
there's
already
a
request
at
a
new
field
from
again
steal
from
a
DM
on
the
object
links.
So
what
is
the
general
procedure
here?
Can
we
already
allocate?
But
it's
not
part
of
this
spectacle.
Specification
is
external.
F
F
Mean
the
salmon
can
be
done
very
quickly
as
soon
as
the
document
is
approved.
You
know,
even
if
it's
not
published
yet
I,
think
that
will
be
okay
to
actually
basically
say
well.
Ayana
will
create
it
right
away.
In
fact
they
usually
typically
create.
You
know
draft
versions
of
the
registries,
or
at
least
they
suggest
tell
you
what
the
registry
is
going
to
contain
during
last
call,
so
they
will
have
information
available
before
they
tell
each
are
typically.
K
So
next
up
fits
and
touch
media
types,
of
course
in
ml.
So
the
background
for
this
one
is
that
you've
been
seeing
what
a
lot
of
lot
of
synonyms
earlier,
but
this
one
part
of
our
use
of
cinema
called
the
collection
use,
and
that
is
what
is
used
today
if
I
became
protocol
and
also
I'm
so
smart
objects.
So
here's
an
example
of
an
in
to
smart
object.
It's
a
remote
actuator!
It
has
actually
a
cure
for
light.
It
has
that
resource
filling
if
lights
on
and
off.
K
K
Use
case
here
is
that
what
about,
if
you
want
to
retrieve
or
change
only
two
of
these
resources,
you
could
be
as
of
today,
doing
two
gets
on
a
specific
resources
and
get
the
values,
and
that
could
be
okay.
How
about?
But,
however,
if
you
want
to
do
this
in
order
to
save
battling
only
with
a
single
request,
or
maybe
you
want
to
be
changing
the
value
of
a
few
of
those,
but
not
touching
attitude.
K
Luckily,
we
do
have
methods
for
that.
In
in
koa
we
had
the
cooperation
patch
or
exactly
for
addressing
a
parts,
resource,
updating
or
fetching
them.
However,
the
KOA
patch
and
fetch
methods
they
do
need
a
payload
format
that
is
dependent
on
the
resource,
creepers
intention
format
and
current.
There
is
no
one
force
nml,
so
that's
exactly
what
is
proposed
by
this
craft,
the
feds
form
loops
last
as
an
example
on
the
bottom
of
the
slide,
it
is
modeled
after
several
JSON
format
axis.
This
is
playing
sin
ml.
K
Without
the
values
and
the
idea
they
recent,
you
can
have
simple
parsing
of
these
type,
a
Log
form.
Of
course,
you
already
have
support
of
sentiment
on
the
device.
It's
using
this
format,
a
really
simple
format,
just
indicate
the
names
and
potentially
times,
if
you're
using
x
in
their
other
values
of
the
records
you
want
to
fetch
and
you
will
be
getting
the
results
for
those.
So
this
would
be
solving
the
problem
of
getting
two
resources
out
of
the
center
go
back
at
Susanna
mode
records
in
its
interval,
words.
K
So
that's
the
simple
part
of
the
best
cuts
format,
but
then
there's
the
optimization
base
now
currently
called
wildcards,
and
essentially
it's
about
when
you
want
to
be
selecting
multiple
records
using
a
single
pitch
or
Pat's
record
again,
it's
if
you
have
a
let's
say
a
few
hundred
people
records
and
you
don't
want
to
be
naming
explicitly
hundred
different
records
on
your
pets
or
Pat's
request.
You
could
be
using
some
kind
of
a
way
to
select
multiple
multiple
records.
A
use
case
could
be,
for
example,
ghetto
temperature
sensor
values
or
be
more
like
10%.
K
So
then,
the
considerations
are
recording
this
track
on
the
wild
cards
is
the
upper
half
on
screen?
Okay,
something
simple!
It's
needed
for
addressing
this
use
case
and
there
was
quite
a
lot
back
up
or
what?
What
is
the
right
way
of
doing
that?
The
current
is
such
that
say.
Early
proposal
are
up
for
discussion.
We
believe
the
basic
format
that,
were
you
explicit
name,
the
resources-
it's
been
a
straightforward,
but
this
world
or
is
something
bit
more
complex
and
maybe
need
a
spark
more
thinking.
K
And
although
what
we
were
discussing
right
before
the
courses
in
here
is
that
there
is
need
for
the
pitch
packs
format
relatively
early,
be
pretty
much
end
of
this
year.
However,
this
wild
card
format
is
maybe
something
we
want
to
consider
a
bit
longer
than
triangle
endpoint
of
this
year.
So
it
could
be
split
out
of
this
document,
because
it
is
quite
a
bit
of
complexity
and
getting
it
right
maybe
requires
some
some
time,
but
the
basic
format
seems
to
be
pretty
straightforward
and
something
what
we
could
be
defining
by
the
end
of
the
year.
B
Sec
shall
be
armed,
telepathic
connection.
So
soon
as
I
stood
up,
you
realize
what
I
was
gonna
say
right,
yeah,
so
I
thought
yeah.
We
always
have
to
be
a
little
suspicious
about
need,
and
it
sounds
to
me
like
this
wild
card
things
nice
to
have
now.
If
you
look
at
how
some
of
these
resource
structures
were
designed
like
in
OMA
objects
right
in
just
referencing
into
like
halfway
down
the
tree,
will
get
you
the
entire
object
in
there.
B
Give
good
recommendation
on
proper
object
design,
so
that
might
be
something
you
you
could
do.
It
say
like.
We
do
not
have
wildcards
in
this
version
of
this
specification.
Thus
here's
some
recommendations
on
how
you
can
structure
your
resources
in
such
a
way
that
this
is
useful
already,
not
right.
That
might
be
a
temporary
make.
K
Sense
and
all
in
all
this
being
able
to
select
multiple
things.
Maybe
it's
what
we
need,
but
a
habit
more
thinking
think
about
the
use
is
probably
and
probably
applies
beyond
10
ml,
so
cinema
is
not
where
we
hit
it
the
first
time.
Maybe
something
even
think
they
are
cheese
called
want
to
think
about
in
a
bigger
picture
and
then
do
later
on,
for
example,
and
extends
with
this,
but
having
the
basic
basic
semantics
or
fetch
and
patch
defined
seems
pretty
straightforward.
K
So
if
we
then
forget
about
the
world
card
considerations
for
a
while
and
yet
there's
also
question
about
regular
expresses-
maybe
it
maybe
later
preferably
not
and
under
the
patch
operations.
So
currently
the
semantics
is
that
you
can
basically
create
new
and
you
can
change
existing
ones,
but
you
cannot,
for
example,
and
a
similar,
so
there
could
be
some
need
for
operation
codes
later.
K
We
can
actually
use
the
same
existing
Golan
format,
IDs
and
justifying
the
semantics
with
the
patch
and
pets
that
are
are
slightly
different
from
the
regular
say,
never
use,
but
then
the
key
question
since
this
now
zero
zero
crap
Eastern
interesting
the
croup
to
work
on
it
because
there's
little
interest
external
positive,
so
OMA
DM
has
already
requirements
for
something
like
this
and
to
me,
I
got
a
bunch
of
others
here.
It
seems
very
sensible
thing
to
do
here,
a
general
thing
that
is
used
by
oma
DM
instead
of
comity
of
specifying
it
work.
K
K
L
M
Mattias
Kovich
Siemens,
so
for
these
wild
cards
there
could
be
a
use
case
on
industrial
systems.
You
very
often
have
this
right,
multiple
read,
multiple
and
so
on.
So
I
would
have
interest
to
to
work
on
what
is
there
that
we
need
I'm,
not
specifically
interested
in
the
cinema
version,
but
in
this
generic
thing,
how
we
could
solve
this
multiple
read
right
sounds.
N
K
K
We
don't
have
a
good
idea
yet
supporting
the
simple
here's,
the
name,
if
you
already
support
getting
senemo
as
input
the
code
change
should
be
pretty
minimal,
but
you
already
have
all
the
machinery
who
understand
that,
and
it
sends
matter
of
speaking
everything
with
map
it
doesn't
explicit
match.
So
I
think
that
should
be
pretty
straightforward
on
the
wild
card
of
filters
and
selectors.
It
really
depends
on
the
design
and
what
one
question
for
it's
on
the
wild
card
was
afraid
how
about
more
than
one
wild
card.
K
O
Is
des
Browne
speaking
about
wild
cards,
one
of
the
things
that
I
was
going
to
say
is
that
wild
cards
in
the
middle
of
the
URI
are
the
or
the
evil
part,
and
you
said:
oh
that's
exactly
what
I
want
so,
but
to
echo
what
Zach
was
saying
so
very
often
with
proper
structuring
of
the
data.
You
can
avoid
that.
O
If
you
think
ahead
to
what
your
use
cases
are,
you
can
avoid
it
and
possibly
if
your
structure
is
something
that
you're
set
with
it's
built
in
stone,
but
you
really
want
to
query
their
different
way.
There's
nothing
wrong
with
having
data
appear
in
different
parts,
different
trees,
different
views
and
so
restructure.
O
The
data
over
there
in
a
different
view,
which
allows
you
to
be
properly
hierarchical
that
it
at
the
very
least,
put
your
wild
cards
only
at
the
end,
because
wild
cards
in
the
middle-
and
we
know
that's
the
search
writing
the
code
for
that
is
very
difficult,
especially
if
you
have
something
that
then
you
open
up
multiple
wild
cards
in
yuku,
you
know
so
it
different
levels,
it
gets
very
complicated.
Wild
cards
on
the
end
are
ok
and
if
you
structure
your
data
so
that
that
works,
that's
ok,
yeah.
K
K
Of,
for
example,
like
with
in
managing
Docs
in
to
smart
object,
you
would
know
okay,
this
object.
Id
can
be
anything.
So
there
are
some
structure
where
it
works,
but
I
do
what
we
in
general.
It
is
thinking
right
and
that's.
It
I
think.
Coming
back
to
you,
or
rather
like
do
another
view,
that's
exact
what
Cristian
was
suggesting
earlier.
Yes,.
P
Alexander
path:
yes,
I'm,
going
to
echo
this
this
comment
and
just
at
a
point
here
that
in
karma
we
are
using
fetch
for
this
couple
of
operation.
So
whenever
you
want
to
do
like
like
basically
what
we
get
as
an
input
is
that
tree
selection
like
a
subtree?
If
you
want
to
go
and
do
some
operations
of
trees,
pretty
pretty
much
everybody
who
is
asking
for?
P
So
maybe
you
know
just
if
you
happen
to
want
a
couple
of
sub
trees,
then
in
a
fetch
operation
you
can
get
them
all
of
the
things
that
you
are
wanting.
I
do
see
the
use
case
where
you
might
want
to
say:
okay,
I,
don't
know
exactly
all
the
identifiers
that
are
here
and
you
know
I
don't
want
to
really
restructure.
If
some
SDO
has
fixed
some
some
particular
you
know
structure
of
your
eyes,
then
you
know
I,
don't
want
to
redo
this,
but
then
I
mean
there.
P
N
A
Would
be
interesting
interested
in
reviewing
future
versions
of
this
document?
1
2
3
4
5
six,
so
we
do
have
some
some
interest
in
this
room
and
I.
Think
I'm
only
going
to
ask
for
formal
adoption
when,
when
cinnamon
is
actually
approved,
to
reduce
the
number
of
documents
in
the
backlog,
but
I
think
we
have
some
interest
here.
Let's
go
ahead
with
us:
okay,.
H
H
Only
the
main
resource
document,
which
is
nearing
completion,
but
also
two
documents
that
are
justly
related
to
it:
the
SSD
document,
which
Carolyn
is
primarily
working
on
and
the
personal
draft
that
moves
some
things
that
we've
been
discussing
in
resource
directory
out
of
the
main
out
of
the
main
document.
H
So,
as
I
said,
the
the
resource
directory
as
it
is
is
from
the
office
point
of
view,
pretty
much
ready,
so
nothing
more
will
come
of
it.
If
we
stay
brooding
over
it.
So
will
we
need
your
input,
so
this
can
go
into
a
working
group.
Last
fall.
We've
done
quite
a
bit
of
work
in
the
last
few
years.
It
is
mainly
documented
in
the
Isha
tract
and
pull
request.
By
the
time
we
released
the
13
draft.
We
were
down
to
one
remaining
issue
which
I'll
talk
about
in
just
moment.
D
H
The
last
days
another
came
up.
That
is
mainly
editorial.
Do
we
want
to
have
this
forward
reference
deal
in
it
or
not,
and
I'm
hoping
for
this
number
to
go
up
a
little
bit
and
while
we
are
receiving
reviews,
but
I
expect
that
there
will
be
mainly
editorial
things
that
we
need
to
clear
out
through
that
time.
H
The
issue
that
I'm
looking
forward
to
solving
is
that
we'd
like
to
show
the
interoperability
of
the
resource
directory,
as
we
have
specified
it
now,
in
that
sense,
that
different
people
arrive
at
compatible
implementations
from
the
same
document.
So
far,
I
have
three
people
that
would
like
to
work
on
this.
Please.
If
you
have
an
implementation
of
a
resource
directory
drop
me
a
mail
and
then,
if
I
know
who
else
is
participating,
then
we
can
set
a
date
and
we
can
set
a
context
and
in
which
we
will
have
this
in
detail.
H
This
will
probably
be
a
remote
interrupt
because
it's
because
we
can
so
what
what
changed
recently.
This
is
basically
the
the
Delta
since
since
I
TF
100
there's
been
lots
of
smaller
cleanups.
So
you,
if
you
want
to
do
this,
then
this
is
exactly
how
it
is
done,
because
the
mother
document
is
not
completely
queue
about
it.
If
you
want
to
indicate
your
server
version
you
can
you
can
do
that
without
changing
anything
of
the
mechanism
we've.
H
We
now
allocating
a
multicast
address,
so
the
initial
step
of
this
finding
a
resource
directory
can
be
targeted
towards
a
resource
directory
without
exactly
knowing
what
it
is.
The
there
big
change
since
12
is
related
to
what
Caston
talked
about
before
the
fine
points
between
RFC
66
90,
resource
link,
format
and
link
headers,
and
for
the
sake
of
not
creating
confusion
and
having
things
just
work.
H
So
when
the
resource
directory
mandates,
that
link
format
is
supported
by
the
server,
but
all
other
kinds
of
serialization
of
web
links
can
be
used
as
well,
and
if
they
don't
suffer
from
the
same
ambiguities,
then
in
those
cases
relative
references,
we
don't
forbid
relative
references,
but
we're
there
where
we
know
there's
problems,
we
just
just
hand
out
absolute
references
and
the
last
part
of
changes
was
related
to
resource
directories
that
are
not
necessarily
hosted
on
the
same
server.
H
H
Another
document
that
is
closely
related
to
the
resource
directory
is
the
is
the
export
from
resource
directory
to
DNS
SD.
This
document
is
still
in
active
work.
It
has
been
updated
to
reflect
also
changes
in
resource
directory.
It
now
has
an
intro
a
proper
introduction,
so
this
this
is.
This
is
not
nearing
completion
yet,
but
it
is
in
a
state
where
we
can
be
relatively
sure
that
we
only
reference
to
the
resource
directory
and
don't
need
to
forward
reference
from
the
resource
directory
to
our
DNS.
H
St
everything
that
DNS
st
is
supposed
to
do
can
happen
as
an
extension
to
the
resource
directory,
as
we
have
it
now,
and
this
is
this
is
the
way
to
go
for
what
we
think
with
this,
and
this
is
happening,
but
not
ready
for
a
working
group.
Last
call,
as
of
today
speaking
of
working
group,
last
call
what
is
missing
for
the
resource
directory
reviews,
because
the
last
once
we've
received
from
Jim
and
another
review
we're
on
11.
H
L
I
see
at
least
three,
maybe
some
of
the
other
NGOs
OCA
or
may
could
also
volunteer,
since
they
use
resource
directory
and
their
members
in
the
room
of
those.
A
A
H
A
A
Yeah,
so
we
have
to
draft
here
that
there
are
working
of
documents
and
one
draft.
There
is
completely
new
and
the
question
is:
who
cares
about
this
de
replication
scenario?
Christian
brought
up
so
do
we?
Do
we
have
the
energy?
Do
we
have
the
interest
in
this
group
to
look
at
this?
Is
this
something
we
should
push
to?
The
research
group.
L
H
Research
research
group
sounds
good
to
me
because
I
this,
this
idea
I,
don't
plan
on
this
to
be
the
way
how
those
things
work,
but
I
like
this
to
document
that
it
that
it
could
work
and
to
explore
what
we
are
running
into
when
we
are
when
we
are
when
we're
doing
this.
So
probably
I
should
just
ask
t2
TRG.
If,
if
there's
interest
in
picking
this
up,
there
yeah.
L
S
It's
too
soon,
I
don't
have
any
strong
opinions
for
or
against
any
solution.
I
do
feel
like.
We
will
want
some
kind
of
solution,
we're
not
there.
Yet
most
homes
have
a
single
Wi-Fi
access
point
and
if
it
fails
it
fails
so
clearly
we
don't
have
a
fault-tolerant
system,
but
I
think
we
will
want
something
like
that
in
the
near
future.
Thank
You.
K
K
A
So
I
think
we're
done
with
this
point
here
Michael.
Since
we
are
ahead
of
schedule,
we
might
be
pulling
the
group
communication
to
today
so
just
prepare
for
that.
But
the
next
item
on
the
agenda
is
come
on
and
I
have
to
dig
out
the
slides
for
a
little
so
start
talking
and
I
will
try
to
find
the
slides
and.
G
P
P
So
there
have
been
several
quite
exciting
things
out
there
and
and
now
the
basic
message
that
is
the
takeaway
from
you
is
that
we
really
need
your
input
now
and
he'll
be
asking
for
all
of
reviews.
So
the
looking
at
all
the
people
that
are
here
each
of
you
can
get
a
documentary
review.
So
that
could
be
good,
but
I
really
am
going
to
be
to
be
asking
for
you
for
this.
So
what
we
done
the
last
time
and
what
we
had
so
we
that
there
are
new
releases,
new
versions
of
all
the
documents.
P
So
there
are
three
main
documents
for
three
main
documents
for
Africa.
My
so
we
have
the
yang
to
see
bore
the
carmine
document
in
the
state
document.
So
these
are
all
working
group
items
and
the
yang
to
see
bore
is,
from
the
point
of
view,
the
altars.
It
is
ready.
So
we
would
like
to
make
a
working
group
last
call
on
this
one.
And
yes,
so
please
do
review
this
one
and.
A
A
P
So
we
did
some
minor
changes
there
to
the
yang
to
see
Boer
and
it
is
ready
for
a
working
group
last
call
so
basically
the
things
that
we
did
there
is
you
have
this
thing
that
is
called
young
template
and
we
just
needed
to
make
sure
it
is
in
rest,
conf
and
we
just
needed
to
make
sure
that
it
works
out
of
the
box
which,
with
the
encoding
with
a
young,
to
savour
encoding
and
as
of
today,
it
works.
So
basically
I,
don't
see
anything
that
we
could
add
to
this
document.
P
P
We
need
to
to
see
if
all
the
things
that
we
added
actually
that
we
made
sure
for
the
yang
to
see
bor
are
actually
also
true
for
the
sit,
a
document
and
go
again
through
a
jana
about
this,
and
I
would
like
for
us
to
make
a
working
group
last
call
after
this
and
on
the
comma
I
draft
well
here
we
need
to
go
over
a
couple
of
things
to
make
sure
that
it
works
out
of
the
box.
The
first
thing
is
the
young
template.
P
Then
there
is
this
new
thing,
or
at
least
for
me
it
was
relatively
new,
the
yang
attached.
We
need
to
make
sure
that
it
works
with
nmda
and
for
yang
push.
So
we
talked
to
during
the
weekend
with
wood
Hank,
so
we
found
out
that
there
is
basically
it
should
work
out
of
the
box.
So
it's
pretty
much
okay,
so
this
is
here.
P
They
continued
building
new
stuff,
that
is
quite
powerful
and
we
are
just
making
sure
that
how
it
actually
works
without
changing
the
thing
that
we
did,
we
did
here
so
I
think
that's
a
pretty
powerful
thing
that
we
are
seeing
here
and
and
yes,
so
these
are
the
three
drafts
here
and
there
is
this
new
draft
that
it's
not
an
inn
of
the
receipt.
It's
in
version
two,
so
it's
stable
since
IGF
98
and
it
is
the
coming
introspection
model.
P
P
We
had
a
virtual
interrupt
at
the
last
hackathon,
so
we
tested
it
with
fetch,
and
so
we
just
the
fetch
operator
without
that
system,
and
it
actually
proved
to
be
quite
challenging
to
have
such
a
big,
a
module
for
for
interrupt
from
the
start.
So
that
was
a
lesson
learned
and
on
this
hackathon,
what
we,
what
we
did
we
scale
down
a
little
bit
on
what
we
wanted
to
do.
We
wanted
to
start
an
open-source
implementation
of
kamae,
so
we
were
postponed
in
this
for
the
next
time.
P
But
what
we
did
do
is
we
joined
the
semantic
interoperability.
We
she
hackathon,
where
it
you
had
the
things
coming
from
coming
from
the
t2
TRG
coming
from
the
w3c,
though
so
you
have
this
thing
description
and
what
we
showed
was
that
you
can
do
the
mapping
from
yang
to
the
thing
description
and
you
you
have
the
bindings
from
comma
to
expressed
in
this
thing
description,
a
lesson
learned
from
there.
P
So
if
you
remember
the
last
tight,
if
you
are
saying
okay,
we
would
like
to
simplify
the
document
and
remove
everything
except
fetch
and
patch
right
now.
This
thing
actually
seems
to
point
out
that
it
is
it's
going
to
only
I
mean
the
whole
day.
The
ecosystem
as
a
whole
would
take
a
little
bit
of
time
to
be
able
to
express
everything
in
only
fetch
and
patch,
so
it
would
seem
a
little
bit
too
fast
to
go
and
say:
well,
we
don't
need
get
input.
P
M
Matthias
Kovac
speaking
for
WC
vapor
things
so
in
the
TD
we
try
to
to
allow
for
adaptation.
So
if
you
have
an
implementation
that
uses,
let's
say
weird
methods
that
we
can
allow
for
that,
however
they
have,
there
was
quite
some
pushback
there
that,
like
the
sensible
default,
should
be
in
place
that
there
is
something
that
people
can
converge
to
and
something
having
a
get
for
something
that
this
link
to
is
is
kind
of
one
of
those.
If
you
want
to
simplify
and
say
hey,
this
is
a
very
constrained
environment.
M
P
So
so
that's
actually
a
very
good
first
lesson
that
that
we
that
we
have
out
of
this-
and
it
was
you
know
at
some
point-
we
were
super
excited,
that's
okay,
when
all
we
can
do
really
super
simplify
the
thing
have
only
fetch
in
patch,
but
it
from
all
this
interoperability
point
of
view
and
from
the
from
semantic
interoperability.
It
wouldn't
be
reasonable.
So
this
actually
says
that
ok,
we
did
our
work
right,
it's
everything's
there.
We
cannot
really
simplify
it
more
than
it
is
so.
P
P
Doing
this
one
of
the
difficulties
that
you
know
we
were
using
ipv6
and
UDP,
so
you
tend
to
have
some
filtering
from
time
to
time
with
these
things
and
we
had
these
issues
so
from
the
last
IDF
we
saw
some
great
people
here
that
work
at
eff
inter
up,
and
so
their
logo
is
cropped
here,
but
they
will
forgive
us
about
this,
so
I've
interrupt
for
those
of
you
who
have
not
heard
this
is
a
platform.
So
it's
an
open
platform,
the
you
can
go
and
you
can
do
your
interoperability
tests
on
that
platform.
P
So
you
have
like
an
environment
that
is
set
up
for
you.
It
is
open
like
it's
a
European
project
like
so
it's
you
know
it's
really
used
for
this
and
you
have
the
way
to
coordinate
the
things
to
just
nip
the
traffic
that
is
so--but.
When
you
do
your,
inter
up,
you
have
a
sniffer
that
is
going
to
write
down
the
all
the
packets
that
are
exchanged
and,
and
you
can
have
even
in
some
cases,
a
verdict
that
will
tell
you,
okay,
that
the
interrupt
passed
or
failed.
P
So
what
we
did
is
we
actually
published
our
implementation
on
on
this
on
this
side,
so
it
is
called
like
a
reference
implementation,
which
is
reference
for
the
eff
interrupt.
So
this
means
that
you
can
go
there
and
you
can
set
up
your
own
interrupt
session
and
you
can
run
it
and
it
will
run
and
you
can
have
come
I
server,
setup
and
akamai
client
so
depending
if,
if
you
wish
to
test,
if
you
have
a
comment,
you
can
go
there
and
test
with
Akamai
server.
P
If
you
have
it
from
my
server,
you
can
use
the
come
by
client,
that's
over
there
and
for
the
moment,
the
methods
that
are
implemented,
our
get
fetch,
put
patch
I
patch
and
delete.
So
we
have
not
yet
implemented
the
post
so
just
to
working
to
show
you
a
couple
of
photos
that
it
really
works.
So
you
go
to
this
Internet
address
you
sign
up
and
then
you'll
get
good
to
go
and
then,
when
you,
when
you
sign
up
the
only
thing
that
you
need
to
you
know
you
say:
okay
I
want
to
new
tests.
P
You
need
to
click
here
on
interoperability,
and
then
you
see
the
because
you
here
you
have
a
list
of
tests
and
you
need
just
to
click.
Ok
I
want
to
do
come,
my
test
shoot,
and
then
you
know
you
click,
Next,
Next
and
and
so
forth,
and
you
have
like
you
have
all
think
out,
because
there
is
a
VPN,
you
can
connect
over
VPN
and
then
you
have
five
to
v6,
and
you
know
that
no
filtering
and
so
forth.
So
it's
really
user
friendly
and
like
it's,
not
us
that
we're
running
it.
P
So
you
have
a
whole
organization
dealing
with
this.
So
really
great.
You
can
start
using
Carmi
today
the
test
file
that
were
published.
So
we
wanted
to
do
something
simplified
and
we
mean
I
Jeff
system.
Is
you
know?
If
you
want
to
make
sense
of
things
it's
it
could
take
a
little
bit
of
time.
So
there
is
a
very
simple
tile
that
it
basically
has
three.
P
So
it's
a
simple
model
with
three
leaves,
so
we
have
a
container
interface
and
three
leaves
IP
address,
name
and
throughput
like
it's
almost
invented,
and
then
we
have
the
seed
file
that
is
generated
out
of
this.
So
you
have
so.
This
is
the
the
identifier
that
number
you
have
comma
int
or
blah
blah
blah,
and
this
is
the
SID.
P
So
it's
as
simple
as
that.
So
that
was
the
first
point
there
is
implement.
So
there
is
the
implementation.
You
can
start
running
it
today.
The
second
point
is
about
the
registry,
so
now
the
the
registry
is
so.
This
is
the
first
draft
of
the
registry.
If
you
wish,
it
is
not
yet
approved
by
a
Jana,
but
we
have
started
already
allocating
ID
so
that
we
can
learn
how
to
use
them.
So
you
can
come
here
so
the
name,
the
URI
Cisco,
my
dot
space
come
I
dot
space.
So
you
go
there.
P
Fortunately,
here
it
is
cut,
but
you
have
a
couple
of
URLs
your
eyes,
so
he
can
go
and
you
can
say
well
what
are
the
more
models
available?
And
so
you
can
see
all
of
them.
So
there
are
some
of
the
system's
ones
are
over
here,
so
you
can
see
the
models
and
you
can
click
on
the
yank
file
and
the
seed
file.
You
know
this
is
only
about
the
sits
right
Yankees.
Just
for
consulting
that
thing
and
you
have
tools
you
can
generate
the
seed
file
from
here.
P
You
can,
you
can
generate
the
egg
file,
so
you
can
do
all
the
other
young
file,
but
you
can
generate
the
city
file
and
all
that
stuff.
So
it's
all
there
just
go
there
and
with
this
I'm
going
to
say,
ok
well,
we
want
we
need
reviews
of
young
to
see
Boer
and
I.
Think
the
best
well,
this
is
just
starter
working
group
last
call.
P
We
have
core
and
the
the
sit
and
the
comma
drafts,
and
here
we
really
need
neat
reviews,
so
we
will
check
these
things
and
you
know
by
the
end
of
March,
that's
sir,
by
the
end
of
March
us
for
sure,
maybe
I'm
not
sure
by
the
end
of
the
week,
we'll
bind
of
by
the
end
of
March
and
and
yeah.
Then
we
need.
We
really
do
need
reviews
I
mean
we,
as
authors
have
done,
have
done
so
I'm,
reflecting
what
Christian
said.
P
T
Hi
this
is
listening.
Half-Year
I
can
make
a
exploration,
endeavor,
Internet,
conf
and
maybe
select
a
specific
individual
there.
That
helps
me
in
concert
to
a
review
on
that,
but
that
is
the
only
way
I
would
commit
to
that.
If
I
do
not
find
a
partner
there,
then
I
will
not,
but
this
I
will
try
to
okay,
okay,.
A
So
the
the
reason
we
haven't
last
called
Yangtze
bow
yet
which
which
is
essentially
done
and
has
been
saving
for
a
while,
is
that
there
is
a
certain
risk,
the
more
we
learn
about
using
it
in
practice,
the
fewer
mistakes
there
will
be
in
there,
but
one
one
view
would
be.
We
can
do
this
now.
We
have
played
with
it
for
long
enough,
even
though
we
don't
have
specific
event
where
we
could
say.
A
A
A
I
forget
what
dimdim
means,
but
it
essentially
means
let's
redesign
everything,
and
so
all
the
the
documents
that
into
H
with
the
Netcom
freeing
space
need
huge
weeks
to
make
sure
that
they
actually
work
in
the
nominal
space
as
well
and
right
now,
I
personally
am
I'm,
not
feeling
very
sure
about
the
fact
that
we
have
done
this,
but
I
also
don't
have
any
indications
that
we
don't
have
done
this.
So
that's
something
where
we
again
need
input
from
the
network
management
people
and
the
same
thing
is
true
for
the
yang
library
I.
A
Think
it's
not
yet
entirely
here
that
we
have
to
do
this
at
all,
but
it's.
It
may
be
a
good
idea
to
have
our
own
way
of
getting
this
catalog
information,
so
it
would
be
really
useful
to
have
input
from
people
who
want
to
use
comai
whether
we
should
be
pursuing
this
or
not
yeah,
so
I
think
as
a
worker.
A
U
A
Okay,
let's
try
this
out
with
young
people.
The
reason
why
young
people
really
would
benefit
from
from
an
early
last
call
is
that
there
are
people
who
want
to
use
yang
as
a
data
modeling
language
way
outside
of
net
Conniff,
just
used
yang
as
a
data
modeling
language
and
and
used
the
CBO
representation,
so
the
this
draft.
It
contains
everything
that
is
needed
for
that.
So
we
can
can
go
ahead
with
those
applications.
A
L
Yep
one
comment
not
related
with
this
one,
but
with
the
DNS
there
we
saw
rectory
DNS
SD
caroling
is
pointing
out
in
the
jabber
that
the
DNS
is
the
working
group
in
on
Thursday.
Will
that
drop
will
be
presented
on
that
group
on
Thursday
for
those
interested
yeah
I
said
at
the
beginning,
Jerry
was
a
reminded.
V
V
Yes,
most
of
the
updates
are
actually
out
of
two
major
reviews.
We
got
from
ESCO
and
Peter
after
Singapore.
Thank
you
very
much
for
that.
Indeed,
so,
based
on
that,
on
those
reviews,
we
essentially
euro
structure,
the
document
organization,
quite
a
lot
and
clarified
as
many
things
are
possible-
and
here
just
summarize,
the
the
main
ones
we
took
care
of
already
started
from
the
terminology.
We
notice
that
there
could
be
potential
for
confusion,
a
third
work
group,
so
we
we
provided
the
exact
definition
we
intend
to
have
for
that.
V
We
refer
to
a
security
group
as
a
set
of
n
points,
sharing
the
same
security
context
and
essentially
key
material
and
policies
and
so
on.
So
this
doesn't
have
to
be
confused
with
other
possible
meanings
like
network
group
when
it
comes
to
possible
IP
addresses
identifying
a
multicast
group,
for
instance,
or
even
application
group,
meaning
a
set
of
nodes
running
the
same
application
in
the
context
of
a
security
or
network
group.
So
here
we
mean
only
security
group
and
you
you
can
map
that
quite
flexibly
in
other
kinds
of
groups
as
well.
V
In
the
security
context,
especially
other
comments
from
Peter
I.
Guess
we
try
to
clarify
as
best
as
possible
how
the
different
types
of
contexts
so
commands
and
the
recipients
are
established
and
arrived
later.
Derivations,
especially
occurring
around
time
and
obvious
requests
as
well.
We
added
a
new
table
highlighting
the
exact
deltas
from
the
contexts
we
already
have
in
the
menos
core
document
and
and
do
the
content
we're
adding
here
in
this
one
in
section
3
about
the
the
cozy
object.
V
6
is
a
new
one
to
have
definitely
because
considerations
on
what
the
group
manager
is
supposed
to
do
in
the
difference
days
of
the
protocol
were
actually
scattered
all
over
the
document,
so
we
essentially
collected
all
of
them.
Now
they
are
list
all
together
in
a
single
section,
where
the
last
two
or
three
are
actually
optional
features
that
the
group
manager
can
help.
So
now
they
are
all
collected
in
a
single
spot.
V
We
had
some
work,
also
in
the
appendices,
actually
that
they
were
a
little
bit
restructure,
rephrase
the
move
appendix
a
used
to
be
a
section
with
assumptions
and
security
objectives,
and
it
was
quite
distracting
where
it
was,
and
it
made
also
cooler
the
alignment
with
the
menos
core
document.
So
we
thought
just
to
move
that
as
an
appendix.
V
We
got
some
feedback
in
those
reviews
about
the
use
cases,
so
we
just
improved
their
description,
appendix
B,
well
in
Appendix
C,
where
we
already
gave
an
example
of
how
a
group
identifier
that
partially
changed
dynamically
can
be
encoded
now.
It
comes
also
with
an
actual
examples,
considering
again
the
constructor
we
already
had
as
prefix
and,
above
then
in
appendix
D.
This
is
also
related
to
what
we
presented
this
morning
in
India's
meeting.
V
Those
are
the
main
points
addressed
in
the
revision.
Now
there
are
a
few
points
open
to
discussion,
for
which
further
reviews
and
feedback
are,
or
an
welcome.
There's
a
point
about
the
Independence
of
the
security
group
from
IP
addresses
altogether
so
far.
We
have
mostly
thought
in
terms
of
a
request
message,
especially
sent
over
multicast
coops
over
a
multicast
IP
address,
but
there
are
some
cases
were
requests,
especially
selectively.
Retransmitted
can
or
should
be
same
even
unicast
both
online.
The
group
identifier
should
be
just
enough
to
retrieve
the
correct
security
context.
W
V
W
V
V
Oh
yes,
this
is
about
the
terminology.
So
right
now
we
have
force
and
stress
quite
a
lot.
The
multicast
were
so
that's
why
the
terminology
multicast
there,
meaning
an
endpoint,
sending
requests
over
multicast
IP
and
that's
it,
but
back
to
the
comment
before.
Perhaps
we
could
just
consider
to
simply
the
the
same
terminology
we
have
in
our
score
and
that's
it
and,
and
that
would
in
turn,
simplify
also
de
the
enrollment
of
new
nodes
in
the
group
cause.
V
It
just
reduces
the
amount
of
roles
they
can
have
and
by
the
way
they
are
just
the
same
again.
So
you'd
have
positive
side
effects
too.
It
should
be
totally
doable
and
finally,
more
on
the
joint
process.
Actually,
as
I
mentioned
also
before
the
appendix
covering
the
process,
is
still
a
high-level
example,
but
is
structure
at
very
high
level
guideline
more
discussion
on
about
how
to
provisioning
and
handling
public
keys
and
I'm,
pointing
at
the
Essbase
approach.
We
we
have
an
SS
the
recommended
way.
V
V
A
Because
I
would
like
it
if
the
interface
that
you
are
assuming
here
is
very
documented,
that's
good!
If
you
want
to
put
another
joint
process
in
there,
but
it's
also
good
for
verifying
the
east
joint
process.
So
it's
clearly,
do
you
find
what
security
of
the
property
is
you
you
expect
from
that
joint
process,
so
I
think
it
is
a
good
thing
to
to
keep
a
clear
boundary.
Okay,.
V
V
Yeah,
we've
already
talked
about
that,
but
it's
essentially
a
related
activity.
Nice
describing
how
the
joint
process
is
performed,
key
provisioning,
including
using
the
ice
framework
and
whatever
profile
of
AAC.
You
need
to
use
to
secure
communication
between
the
joining
endpoint
as
the
client
and
the
group
manager
as
the
resource
server,
but
you
find
more
details
in
the
ACE
draft
about
that.
A
Okay,
so
generally,
who
is
interested
in
reviewing
this
document,
so
we
have
had
a
few
people
who
reviewed
it?
Okay,
so
there
are
about
five
hands
up
now.
Okay,
thank
you!
So
we
still,
if
I
left-
and
we
could
just
continue
with
the
agenda,
which
would
make
re
the
next
guy.
Who
is
yes,
but
it's
less
juggling
if
we
just
use
them
the
next
item
on
the
agenda,
so
we
have
a
few
surprises
in
store
today.
K
Okay,
so
too
many
requests
response
code
for
code,
how
to
check
what
is
actually
on
the
agenda.
So
the
background
on
this
is
this
common
cold
case
that
a
client
can
be
bombarding
a
server
so
hard
that
the
X
becomes
overloaded,
and
the
question
is
like
how
can
server
tell
hey
you
are
doing
this
too
fast,
please,
back
off
I
nee
to
DP.
There
is
already
find
an
error
code,
four
to
nine
called
tool,
a
requests,
so
it
seems
quite
sensible
to
also
register
it
is
for
coop
with
a
four
point.
K
Two
nine
and
the
acog
specific
edition
here
would
be.
I
would
use
the
max-age
option
to
indicate
for
the
client
when
is
it
okay
to
try
again?
So
that's
rough,
the
gist
of
it-
and
this
is
originally
part
of
the
co-op's
a
broker.
So
it
was
decided
in
the
last
IDF,
meaning
that
it
changed.
This
kind
of
an
error
code
seems
have
used
much
more
wider
scope
than
just
a
broker.
It
makes
sense
to
do
it
as
a
separate
RFC
and
there's
also
a
recent
ocf
interest
for
the
same
kind
of
functionality.
K
There's
all
one
question
that
came
up
during
the
drafting
of
the
details
of
this
specification
is
that,
should
there
be
a
way
for
the
server
to
say?
Well,
don't
do
that,
but
here's
a
set
of
things
that
I
might
be
considering.
Okay
for
you
to
do
so.
What
the
current
text
says.
The
client
shouldn't
sent
the
same
request
before
the
max
of
max
age.
Option
has
passed
so,
however,
something
may
be
useful
here,
but
maybe
again,
this
is
something
wire
and
just
for
this
response
code
could
be
actually
applicable
for
other
response
code.
K
A
Of
it,
our
previous
slide,
so
constant
again
from
the
floor,
this
sounds
like
a
really
good
place
to
remind
people
that
they
can
actually
send
response
bodies
with
error
responses,
and
you
could
explicitly
point
out
that
that
the
array
response
that
comes
back
might
give
additional
guidance,
and
then
people
can
just
register
media
tribes
that
they
provide
that
their
guidance,
and
we
have
a
nice
decoupling
from
respond
fault
code
itself.
Exactly,
however,.
K
Maybe
that's
not
something
you
want
to
specify
in
this
draft
that
has
a
generic
mechanism,
so
I
would
recommend
keeping
that
I'll
scope.
For
this,
however,
would
be
very
useful
to
progress
this
draft
relatively
fast
because
of
its
gonna
be
now
a
normative
dependency
for
the
Bob's,
a
broker
and,
as
it's
also
interest
in
outside
of
the
IEP
f-for.
Using
this,
so
keeping
it
simple
and
being
able
to
progress
seems
like
a
sensible
way
forward.
K
However,
there
has
been
a
discussion
that
well
there's
also
another
and
some
other
response
codes
that
we
may
want
to
be,
defining
and
shoot.
This
draft
then
be
bundled
with
some
of
the
other
other
response
code.
If
they
are
something
that
we
can
progress
fast
and
in
a
non-controversial
matter,
probably
it
makes
sense
if
it
would
be
delaying
publication
of
this
draft.
K
I
would
be
bit
worried
of
that,
but
the
key
question
yeah
okay
working
group
item,
since
it's
a
normally
dependency
and
was
agreed
on
the
last
IETF
that
it's
already
it
was
already
part
of
a
group
document,
but
we
just
played
it
apart.
Having
work
group
working
on
this
seems
to
be
a
reasonable
way
forward.
W
Peter
farmer,
stock
I
also
had
an
controversial
response
code,
which
has
been
presented
already
twice,
I
think
and
I
very
much
would
like
it
to
go
forward
it
the
same
smooth
way
as
is
proposed
for
this
response
code
and
I
would
say
that
bundling
the
two
response
code
to
have
the
same
smooth
treatment
would
be
very
welcome.
Indeed,
you
know
I.
A
But
let's
do
this
this
thing
first,
so
anything
else.
We
need
to
discuss
about
that.
So
you,
you
have
a
little
bit
of
feedback.
You
can
do
a
dash
to
zero
ITF
the
view
of
the
document
after
the
Wagner
adoption,
and
then
we
do
with
the
working
class
code
and
if
Peter
is
fast
enough
to
get
his
stuff
in
there
before
the
way
you
bla
scroll,
immunes.
Oh
the
cheese,
okay,.
A
D
D
W
Okay
and
please
to
oblige
what
is
the
motivation
for
this
draft
that
the
robot
of
secure
transport
uses
the
HTTP
response
202
and
which
means
it's
not
immediately
available
but
say
come
back
in
ten
minutes
three
hours
and
then
you
can
get
the
response
and
this
response
code
can
be
returned
both
for
and
get
or
on
post,
so
just
doing
an
observer
on
a
ket.
It's
not
sufficient,
because
you
want
also
on
the
post
to
have
this
response
caused
though
we
have
no
such
response
code
for
co-op
and
nevertheless
we
need.
W
W
Well,
we
have
three
use
cases
for
this
kind
of
response
codes.
First
of
all,
there
is
this
is
D
co-op
s.
Then
there
is
the
cutoff
score,
perp
SERP,
which
has
a
bit
different
treatment.
It
doesn't
say
you
have
to
wait
for
the
raise
for
the
result,
but
actually
it
says
there
is
no
result
yep,
so
you
go
away
and
don't
bother
me
and
then
there's
of
course,
the
too
many
requirements
which
we
have
from
we
have
just
about,
which
is
also
a
new
response
code.
W
A
So
basically,
co-op
is
is
a
lot
like
HTTP,
except
where
is,
and
we
have
tried
to
keep
the
the
color
state
machine
simple.
So,
for
instance,
we
have
decided
not
to
do
redirects,
even
though
redirects
are,
but
by
mini
HTTP
people
consider
to
be
a
basic
part
of
HTTP
that
everybody
should
always
do.
Koib
doesn't
use
them,
and
there
are
two
reasons
for
this.
A
One
is
redirects
are
really
badly
defined
in
HTTP,
so
that
that's
would
be
hard
to
do
this
right,
but
also
we
just
want
to
keep
the
state
machine
Center
and
the
difference
between
error
response
codes
and
success.
Response
codes
is
that
error
response
codes
all
create
the
same
imprint
in
the
state
machine.
There
was
an
error,
so
the
request,
as
done
no
caching
and
so
on,
while
a
success
response
code
actually
requires
every
single
client
to
have
code
to
handle
this
success
case
now.
A
One
one
way
to
handle
it,
of
course,
is
to
say:
I,
don't
understand
this,
so
I
give
up,
but
okay
I
think
that's,
that's
really
counterproductive.
So
adding
a
success
response
code
is
an
expensive
operation
because
kind
of
it's
the
whole
ecosystem,
and
now
that
the
question
is,
if
there
are
application-specific
state
machines
that
that
happened
to
have
something
in
HTTP.
That
kind
of
seems
to
have
handled.
That
I
mean
that
the
way
ESD
has
appropriated
tort
words
is
I'm
not
going
to
comment
about
that.
But
is
that
a
reason
for
us
to
emulate
that?
A
Or
should
we
just
say
there
is
a
rest
approach
to
do
these
state
machines
if
an
application
needs
a
state
machine
and
you
transfer
representations
with
the
important
information
to
run
the
state
machine.
So
my
feeling
is
just
as
with
redirect
where
we
told
ocf.
Oh,
you
can
define
a
media-type
that
has
all
the
information
that
that
would
be
an
redirect.
We
could
do
this
here
and
say:
oh,
we
can
defined
a
media-type
that
has
the
content
is
not
yet
available
now.
A
Those
applications
that
that
need
to
cope
with
this,
of
course,
now
have
to
implement
that
media
type.
So
we
have
reduced
the
complexity
to
zero,
but
other
clients
that
they
don't
care
about.
Applications
like
this
don't
have
to
take
an
impact
on
on
their
state
machine
and
I
think
this
is,
since
we
have
about
40
seconds
left.
This
is
one
thing
that
that
you
might
want
to
take
home
for
for
the
afternoon
and
we
can
continue
discussing
tomorrow
morning,
but
I
really
like
to
hear
what
what
the
view
of
the
working
group
on
this
is.