►
From YouTube: IETF-CORE-20220914-1400
Description
CORE meeting session at IETF
2022/09/14 1400
https://datatracker.ietf.org/meeting//proceedings/
B
A
A
D
A
You
really
want
to
join
the
Earth
48
archive
mailing
list
or
you
may
want
to
subscribe
to
it,
not
join
to
it
and
look
for
the
discussion
90
to
90,
because
really
this
is
just
examinator
RFC
stupidity
which
doesn't
allow
us
to
do
the
right
thing.
So
we
are
putting
work
around
on
top
of
work
around
and
I.
D
B
B
B
A
Okay,
you
should
be
seeing
my
slides
now,
yes
good,
so
there
are
two
items
on
the
agenda
and
and
the
first
one
really
is
a
single
slide,
because
it's
more
more
of
a
discussion
point
the
discussion
that
has
been
ongoing
for
a
couple
of
months
now
and
and
that
we
really
need
to
find
an
answer
to,
and
the
other
is
actually
about
document
the
the
href
or
CI
document.
So
let's
talk
about
the
discussion
Point
first,
so
this
is
about
a
document.
Oh
I,
don't
didn't
put
the
document
name
on
this
slide.
A
This
is
about
a
document
that
has
been
created
in
the
anima
working
group
and
that
essentially
is
completed
there.
So
we
we
have
to
make
up
our
mind
quickly
on
on
this
one
and
they
are
providing
a
joint
proxy
mechanism
and
the
the
important
issue
about
joint
proxies,
of
course,
is
yet
that
you
need
to
be
able
to
find
them
and
the
the
current
document
has
different
ways
of
finding
joint
proxies,
and
one
of
these
ways
is
based
on.
Resource
directories
are
multicast
access
to
where
known
core.
A
A
So
the
the
draft
actually
follows
this
and
uses
two
different
resource
types
for
two
two
different
kinds
of
joint
proxies,
which
probably
are
not
that
interesting
to
us
at
the
moment
what
those
kinds
are,
but
the
the
interesting
observation
is
that
a
joint
proxy
is
not
really
a
resource.
It's
actually
a
protocol
that
that
you
can
talk
to
a
specific
server
to
use
that
server
as
a
very
specialized
proxy
to
to
a
join
server.
A
D
A
So
this
is
not
really
further
relevant
to
to
the
discussion
anymore
because
we
are
not
using
TCP
here.
This
is
definitely
a
UDP
based
protocol,
but
the
the
fact
that
people
have
been
making
up
fantasy,
UI
schemes
like
like
TCP
colon,
shows
that
that
it
is
often
useful
to
use
the
UI
syntax
to
to
provide
transport
address
information.
A
D
A
A
D
B
B
C
Ahead,
I
think
it's
useful
I
I
do
agree
that
it's
useful
to
have
this
information,
some
Somewhere
In
Your
Eyes,
but
we
might
even
consider
so
what
we'll
have
to
do
I
think
is
split
at
least
mentally.
What
is
a
statement
about
a
particular
resource
and
what
is
the
statement
about
how
this
resource
can
be
reached,
even
though
it
might
be
expressed
in
the
same
resource?
The
transport
indication
graph
has
terminology
for
that.
C
I
think
we
can
use
that,
but
applying
it
to
co-op
S
Plus
JPY
is
a
bit
trickier
because
it's
not
really
a
co-op
protocol,
but
it's
more
like
a
UDP
wrapper
protocol
and
no
end
device
ever
talks
Co-op
directly
over
that
protocol.
So
it's
it's
a
bit
of
an
in-between
thing.
That's
probably
a
hack
used
because
one
is
about
using
Osco,
but
that's
kind
of
the
point
of
the
draft.
C
A
E
A
Well,
the
the
issue
is
that
that
there
is
no
proper
resource,
so
actually,
in
the
end,
when
this
joint
proxy
protocol
is
is
being
used,
there
is
a
proper
resource,
but
that's
offered
by
somebody
else
and
what
the
joint
proxy
does
is.
It
provides
your
way
to
access
that
resource
that
is
provided
by
some
some
other
sub.
A
It's
providing
the
protocol
parameters.
You
need
to
run
the
joint
proxy
protocol
to
talk
to
somebody
else
and
I
agree.
It's
a
weakness
that
this
entry
doesn't
reveal
to
whom
you
actually
would
be
talking,
but
the
the
other
question
is
whether
actually
this
is
desired
feature
and
if
it's
not,
we
shouldn't
should
force
people
to
do
that.
E
C
Yes,
I
just
want
to
read
from
that.
I
agree
with
with
how
what
you
said
about
that.
It's
that
sorry,
okay,
sorry,
let
me
rephrase
I
think
that
the
way
that
you
mentioned,
that
top
of
talking
about
the
specific
resources
is
the
way
to
go
here,
discover
the
resources
that
are
actually
that
it's
actually
about
and
then
at
metadata
about
how
to
reach
them.
A
A
E
A
A
B
A
Yeah
I
think
the
the
interesting
observation
here
is
that
the
way
it
is
being
proposed
in
the
current
animal
draft,
the
the
client
that
gets
this
resource
simply
doesn't
have.
The
information
who
he
would
be
speaking
to
which
maybe
in
in
certain
closed
environments
is,
is
okay,
but
from
a
general
point
of
view
that
that's
a
bit
weird.
A
A
B
B
C
I'm
not
sure,
because
I
as
I
understand
the
joint
that
that
joint,
that
that
JPI
URI
is
not
the
joint
proxy
itself,
because
the
joint
proxy
itself
is
the
the
the
other
constraint
device,
but
that
your
eye
refers
to
just
another
port
on
the
on
around
the
joint
resource
where
Co-op
s
and
that
joint
protocol
are
baked
together.
But
I
think
that
would
typically
be
on
the
same
host.
Even
no.
A
No
you,
you
really
would
be
looking
for
for
a
joint
proxy.
So
if
you,
if
the
the
pledge-
which
is
their
word
for
the
newly
joining
client,
if,
if
that
finds
a
joint
server,
I
forget
what
the
name
in
in
their
attorneys.
If,
if
the
pledge
finds
a
joint
server,
then
the
pledge
is
all
set,
but
if
it
doesn't,
then
it
has
to
start
looking
for
a
joint
proxy
and
the
The.
A
E
Yeah
just
throwing
it
out
there
in
case
it's
obvious
I
guess
you
could
have
hosted
by
relation
like
on
the
first
proxy,
hosted
by
relations
I
guess
they
also
work
for
different
interfaces
on
the
same
host,
but
also
two
different
endpoints
altogether
Christian.
You
probably
have
thought
about
this
already.
C
B
A
E
A
A
So
there
are
two
issues:
43
and
44,
which
are
both
about
percent
encoded
text,
and
we
we
noticed
that
unreserved
characters
such
as
period
normalized
to
to
being
a
non-present
encoded.
So
we
had
to
make
some
some
changes
there,
and
also
everything
that
is
ugf
age
and
not
in
the
first
128
code.
Positions
also
becomes.
A
It
takes
sorry
if
we
find
a
percent
encoded,
a
text
that
would
not
be
a
valid
utf-a
encoding
of
an
unreserved
Unicode
scalar
value.
Then
we
put
one
byte
in
into
a
byte
string
for
that
preset
encoding
and
repeat
so
we
have
a
well-defined
normalization
algorithm
and
the
the
examples
on
this
slide
is
the
word
globier
circuitry,
which
is
blueberry
marmalade,
and
that
would
just
be
encoded
as
utf-a
text,
because
both
the
the
ASCII
and
the
Beyond
as
key
characters
are
proper
Unicode
scalar
values.
A
A
Oh
interesting,
the
the
observations
is
that
co-ab
requires
net
Unicode,
so
there
are
some
some
additional
requirements
like
no
bums
and
so
on,
but
we
would
not
be
mirroring
those
and
Christian
noticed
that
the
dc47
would
be
just
a
DC
and
then
47
is
is
actually
a
village
character
because
it's
capital
f.
So
excuse
me
capital
G,
so
it
would
be
in
the
string,
together
with
the
U
zero
zero
zero
one.
A
So,
to
summarize,
we
we
my
proposal
here.
Is
we
Define
the
normalization
algorithm?
We
require
that
normalization
algorithm
to
be
used.
This
is
not
Unicode
normalization.
Maybe
the
word
normalization
needs
to
be
replaced
for
something
else.
This
is
just
a
defined
way
to
get
partially
percent
encoded
URI
into
the
patch
structure,
and
we
would
do
this
with
regard
to
utf-8
and
unicode
scalar
values,
but
not
with
regard
to
the
additional
coab
requirements
on
net
unicode.
C
I'm
happy
with
the
state
of
merch
of
much
things
here,
I'll
just
like
to
point
out
that,
because
when
we
take
this
route
of
of
ignoring
that
Co-op
statement,
we
might
be
pushing
co-op
in
whatever
kind
of
later
development
it
has
towards
dropping
that
requirement,
because
there
is
no
practical.
There
is
no
practical
way
for
for
users
to
process
a
URI,
a
CRI
that
is
not
just
taking
the
text
values
into
Co-op
and
failing
if
there
is
any
any
pet.
C
A
A
No,
if
there
is
no
no
bom,
then
we
are
fine.
This
is
really
about
the
the
situation
where
somebody
evil
put
a
bom
into
a
URI
and
then
a
co-op
implementation,
which
is
is
implementing
the
net
Unicode
restriction.
I,
don't
think
such
a
an
implementation
exists,
but
in
Syria
it
would
be
possible.
What
stumbled
about
that.
So
it's
a
bit
of
a
theoretical
case,
but
it's
good
that
we
have
been
thinking
about
that.
A
So
the
the
next
issue
was
that
URI
with
a
scheme,
No,
Authority
and
no
path
actually
could
be
either
considered
to
have
no
path
component
or
could
be
considered
to
have
no
leading
slash
and
then
no
path
component.
So
we
we
have
an
ambiguous
translation
of
a
colon
into
either
just
a
CI
with
a
scheme
or
CI
with
a
scheme
and
then
the
indication
that
we
have
no
leading
Slash
and
we
could
pretty
much
toss
a
coin.
A
C
A
A
B
A
Yeah
I
think
that
that's
a
good
requirement
and
since
I
haven't
implemented
this
yet
I
I,
don't
have
a
strong
feeling
on
how
to
put
this
in,
but
in
in
an
Ideal
World.
Of
course
it
would
be
possible
to
say
this
in
the
cddl,
but
this
is
such
a
context,
sensitive
thing
that
that
cddl
is
a
little
bit
not
strong
enough
to
do
this.
A
A
A
B
A
A
So
we
would
have
the
test
vectors
and
some
some
supporting
softly
in-depth
repository,
which
would
make
it
a
bit
easier
for
people
to
make
use
of
the
test
vectors
that
that
would
be
my
proposal.
So
we
would
have.
We
would
still
have
a
few
examples
in
the
RFC,
but
not
with
the
the
intention
to
cover
all
bases
but
just
to
to
support
the
explanatory
text.