►
From YouTube: IETF-CORE-20221207-1500
Description
CORE meeting session at IETF
2022/12/07 1500
https://datatracker.ietf.org/meeting//proceedings/
A
C
B
B
B
Yeah
Francesca,
if
you're
talking,
we
can
hear
you
on
mobile,
okay.
A
B
Do
that
and
I'll
one
welcome
to
the
interim
meeting
of
the
co-working
group.
It
should
be.
The
last
word
is
here:
this
is
Marco
Michael
chosen
Jimenez
and
custom
Borman,
and
as
a
reminder,
this
is
an
official
meeting.
So
the
note
12
applies.
So
please
note:
well
it's
not
just
about
APR.
It's
also
about
our
conduct,
so
be
nice
and
professional
with
each
other,
and
that
said,
we
have
two
main
items
in
the
agenda
for
today.
I'll
try
to
take
notes
myself.
Any
help
is
appreciated
on
our
notes.
B
The
two
items
are
the
status
I'm
planning,
next
steps
for
the
core
seed
document
from
the
corecon
cluster,
and
we
were
discussing
before
already
that
a
design
team
meeting
is
planned
in
somewhere
in
the
next
few
weeks.
We'll
have
a
doodle
about
that
a
and
then
we
agreed
already
the
previous
interview
meeting
to
wrap
up
on
the
resolution
of
the
Errata
for
the
core
documents,
and
we
got
some
input
today.
I
can
see
from
from
Klaus
on
the
notes
themselves,
and
I
also
saw
a
mail
from
escort
on
the
list.
B
A
Great
yeah
I
actually
have
a
slight
addition
to
the
agenda.
Maybe
we
can
talk
about
the
the
other
car
constructurement
comma
that
we
have
to
get
fixed
soon,
where
we
can
have
a
couple
of
minutes
each
about
the
target
attributes
and
the
Corrections
and
clarifications.
A
So
if
nobody
has
a
problem
with
that
and
then
let
me
go
into
the
car
conf
part
so
call
conf
is
for
documents,
one
one
is
an
RFC
one.
It
is
in
isg
and
and
has
had
a
discuss
for
one
and
a
half
years
now
one
has
passed
working
with
last
call,
but
there
are
some
some
ongoing
discussions
and
we
want
to
make
sure
that
when
we
ship
it,
it
actually
is
up
to
date
with
the
changes
that
the
first
two
documents
took
in
the
isg
process.
A
And
finally,
we
have
the
Yang
Library
document
that
has
a
bit
of
a
problem
in
that
the
the
people
who
really
need
this
haven't
recently
contributed
much
to
it
so
yeah,
but
that
that
will
not
be
the
subject
today.
Let's
first
finish
the
first
three
so
that
that's
the
context
on
call
Sid,
we
have
a
discuss
which
was
from
July
but
not
July
this
year,
but
yeah
last
year.
A
The
remaining
main
issue
in
that
discuss
was
that
we
wanted
to
document
the
objectives
of
the
Sid
management
better,
so
so
the
the
reader
doesn't
have
to
guess
this
from
from
the
mechanism
we
describe
but
actually
gets
the
objectives
in
a
clear
statement:
Dash
19,
which
is
from
July
this
year,
addressed
this
at
least
I
think
this
is
pull
request.
146.
A
I
was
going
to
merge
this
right
after
we
got
feedback
on
that
which
hasn't
happened
yet,
and
I
have
not
been
very
successful
in
in
talking
to
Rob
about
that.
I
met
him
at
iitf
115,
but
we
actually
didn't
manage
to
set
up
an
actual
meeting
about
that.
So
this
is
one
item
where
we
are
essentially
waiting
for
Rob's
input,
but
that
of
course
doesn't
mean
that
we
cannot
have
input
from
the
working
group
on
this
as
well.
So
please
read
that
section
that
new
section
in
dash
19.
A
and
the
the
other
thing
that
came
up
is
that
in
the
workflow,
when
you
have
a
sid
file
that
is
already
merged
from
a
stable
and
an
unstable
input.
So
let's
say
you
have
a
published
RFC.
You
have
a
pretty
stable
draft,
it's
not
yet
stable
in
the
Sid
file
sense,
but
it
it's
nearing
completion
and
then
you
start
adding
two
two
or
three
more
further
items.
A
You
have
the
situation
where
you
have
a
set
fire
that
comes
from
the
Stables
source
of
the
the
old
published
RFC,
the
unstable
source
of
the
current
Internet
draft,
and
you
want
to
use
that
as
input
to
a
further
set
generation
from
the
current
Yang
module
and
that's
not
something
that
that's
currently
supported
by
by
peaking
and
so
far
the
workflow
is
based
on
on
the
the
idea
that
a
whole
file
is
either
stable
and
unstable,
and
that
obviously
doesn't
work
in
in
the
specific
use
case
that
I
just
mentioned.
A
So
we
came
up
with
a
stable
field
which
can
be
true
or
false
defaults
to
True,
which
is
a
bit
weird,
but
yeah
that
happens.
When
you
introduce
a
new
field
late
in
the
process,
then
the
the
old
meaning
has
to
be
the
the
default
and
yeah.
Before
we
really
can
push
this.
We
probably
need
to
find
someone
who
can
support
this
npn
so
that
that's
another
open
flank.
We
have
right
now
at
what
I
mean.
A
This
is
Trivial
enough
that
that
we,
we
are
probably
going
to
get
this
right
without
implementation
support,
but
it's
certainly
better
to
to
have
it
supported
before
we
actually
complete
this
this
document,
so
the
the
whole
set
of
changes
is
probably
large
enough
that
we
will
want
to
do
a
second
working
last
call
on
this.
A
That's
not
for
me
to
decide
that's
for
the
the
other
chance
to
decide,
but
that
would
be
my
my
guess
what
what
the
chairs
will
decide.
So
we
will
go
through
the
whole
process
once
more,
but
I
think
that
that's
a
reasonable
thing
to
do
if
we
think
this
can
be
done
reasonably
fast.
A
So
just
one
one
more
slide
here
with
the
addition
to
the
Sid
file.
So
this
is
a
new
Leaf
and
which
you
put
into
the
description
of
a
specific
Sid
or
the
data
for
a
specifics.
It,
and
this
is
an
numeration
that
inside
a
stable
or
unstable,
with
radius,
0
and
video
one
and
the
default
is
stable.
A
So
that's
the
the
status
of
Sid
we
just
immediately
before
the
meeting
said.
We
wanted
to
have
a
a
design
team
meeting
to
discuss
any
further
questions,
particularly
about
workflows.
I
mean
this
requirement
for
a
stable
field
that
came
up
when
we
were
discussing
workflows
and
we
lost
Francesca.
Apparently.
A
Yeah
media
streams,
don't
always
work
with
mid
Eco
I,
don't
know
why
anyway.
So
this
is
what
what
we
have
done
or
nearly
done,
and
we
probably
need
another
editorial
around
after
the
latest
changes.
A
So,
for
instance,
we
swapped
out
Yang
data
and
and
put
in
SX
structure
instead,
but
there
are
still
some
remnants
of
gang
data
and
and
references
to
the
the
netconf
draft
that
are
no
longer
needed
at
this
point
and
there's
the
usual
checking
you
need
when
you
you
work
on
a
young
module,
you
have
to
look
at
the
potential
changes
that
need
to
be
made
and,
and
there
needs
to
be
a
new
date
in
there
and
we
have
to
check
whether
the
author
said
is
still
right
and
one
other
thing
that
we
probably
want
to
do
is
do
another
review
against
the
comma
requirements.
A
So
this
is
the
the
plan.
I
think
that
the
the
oops
that
is
called
my
I
think
that
what
we
need
to
do
is
pretty
clear.
We
are
waiting
for
this
one
elephant,
which
is
the
discuss,
and
it
would
be
good
to
clear
this
discuss
before
we
go
into
another
round
of
last
calling.
B
B
A
But
again,
that's
for
the
working
of
just
to
decide
whether
the
changes
are
heavy
enough
that
we
should
do
that.
So
my
reading
is
we.
We
do
have
some
changes
here
and
that's
the
the
that
might
be
the
appropriate
thing
to
do.
A
A
Apparently
so
I
I
have
I,
don't
have
like
talking
to
him.
So
I
think
he
hasn't
looked
at
zulub
for
two
weeks
and
he
didn't
get
apparently
didn't
get
my
my
email,
so
yeah
little
nudge
would
be
nice.
A
Yeah
it's
tedious
I
mean
we
can
try
to
to
to
just
slip
by
all
this.
If
Rob
agrees
with
that,
then
I
would
be
willing
to
to
confide
in
doing
something
like
that.
A
A
A
So
normally,
when
you
do
commai,
you
would
use
when
you
do
comma
with
Co-op,
you
would
use
the
fetch
method,
so
you
can
specify
exactly
what
the
parameters
for
for
the
operation
are,
but
there
is
also
a
get
version
of
that,
and
that
of
course
requires
representing
all
kinds
of
information
in
the
your
eye
and
now
why
doesn't
this
Advance
the
slide?
A
Oh
it
does
so,
of
course,
there
are
a
number
of
young
data
types
and
what
coma
currently
does
is
it
defines
its
own
representation
of
these
Yang
data
types
which
sometimes
use
the
the
Yang
sibo
representation
but
sometimes
also
are
different,
and
we
found
that
that
this
was
sometimes
well.
First
of
all,
it's
just
complexity
because
you
suddenly
have
to
have
two
different
representations
of
the
same
information,
but
it's
also
surprising,
for
instance,
that
when
you
go
from
from
an
unsigned
integer
to
a
signed
integer,
the
representation
changes
completely.
A
So
the
unsigned
representation
is
the
decimal
number.
Let's
find
a
representation
is
a
base64
encoded,
sibo
encoded
number,
so
it
looks
completely
different
and
yeah.
There
are
even
some
some
small
editorial
problems
here:
decimal
64
the
entry
has
to
be
written
differently,
but
that
can
be
easily
fixed.
The
problem
really
is
that
this
is
all
different
and
in
in
most
constrained
applications
you
are
not
running
the
gamut
of
of
these
11
different
classes
of
young
data
type.
A
So
this
this
is
just
a
lot
of
effort
you
have
to
do
for
for
supporting,
get
which
you
probably
don't
really
want
to
support
anyway,
because
you
really
really
want
to
work
with
fetch.
A
So
what
we
came
up
with
when
we
discussed
this
at
the
the
hackathon
and
the
itf-115
hackathon
is
to
to
always
use
URL,
save
basically
file
of
the
sibo
of
the
encoded
c
bar
the
representation
of
of
the
key.
This
has
the
advantage
that
it's
uniform.
This
has
the
disadvantage,
that
there
are
certain
cases
when
this
is
longer.
There
are
also
cases
when
this
is
shorter,
so
a
decimal
representation
of
a
larger
number
will
always
be
longer
than
a
base64
representation
of
the
same
number.
A
So
yeah
this
this
trades
complexity
against
yeah,
maybe
the
occasional
additional
bite
being
used
in
a
query
parameter
but
again,
if
you
really
care
about
this
use,
Fetch
and
then
don't
use
okay,
so
the
the
solution
is
is
pretty
clear,
but
if,
if
there
is
a
good
reason
to
use
git,
you
might
be
spending
a
bite
or
two
more.
A
The
the
one
place
that
is
probably
most
frequent
for
many
applications
is
the
string
case,
and
here,
as
you
can
see
from
from
the
table
the
text
string
case,
the
the
old
coma
representation
is
particularly
efficient
because
you
can
just
put
the
the
actual
value,
at
least
if
that
doesn't
contain
an
ampers
enzyme
which
contains
an
FS
enzyme.
Then
you
have
to
do
something
about
that
and
well
that's
in
the
UI
and
they
actually
Co-op
you,
you
don't
have
to
to
cope
with
it.
A
Yeah
has
a
comment
to
to
that
that
describes
this
in
modern
here.
Let's
do
the
exact
wording
for
the
thing
offline
I'm
just
trying
to
give
a
high
level
view
here
so
for
for
the
frequent
case
that
that
we
have
a
string,
we
might
want
to
make
a
special
a
case
out
of
that
I'm,
not
sure
that
it's
worth
doing
that,
but
it's
at
least
so.
If
people
really
care
about
the
the
texturing
case,
we
could
do
that.
A
So
we
would
reduce
the
11
cases
to
two
instead
of
one
which
I
think
is
still
okay,
so
I
don't
know.
If
anybody
in
this
meeting
has
an
opinion
on
that,
but
that's
definitely
something
we
would
need
to
discuss.
But
of
course
the
the
bigger
question
is.
A
Do
we
want
to
make
a
relatively
large
change
here
to
to
simplify
things,
I'm,
always
a
big
fan
of
simplifying
things,
while
I
still
can
simplify
it,
but
of
course,
if,
if
there
are
large
existing
installations
that
use
this
mechanism,
that
would
be
a
little
bit
unfortunate.
A
C
C
Yeah
so
from
yeah
iPhone,
working
implementation
and
implementing
both
get
and
fetch,
as
described
like
to
get
as
described
with
the
different
encodings
currently
in
the
draft,
I
have
to
admit
that
the
different
encoding
is
mostly
complexity,
but
in
terms
of
code
size,
the
different
encodings
doesn't
really
matter
that
much
digging
my
own
grave
hair
in
terms
of
the
proposed
change,
I
would
say
it's
mostly
like
getting
this
consistent.
C
There
was
also
the
edge
case
crushed
and
pointed
out
that
he
had
got
told
that
if
there
is
like
this
is
I
think
currently
a
common,
comma
separated
string
with
different
K
parents.
If
there's
a
comma
in
your
string
key,
then
the
whole
encoding
breaks
but
yeah,
not
much
more
to
add
here,
I
think
I
would
say
going
with
the
or
else
safe
basics
for
encounter
c
bar
of
everything
is
the
safe
way
to
go.
A
A
So
let
me
just
quickly
talk
about
two
documents.
One
is
the
target
attributes
which
we
adopted
as
a
working
group
document
and
I.
Don't
think
there's
really
that
much
to
be
done
there,
but
we
probably
want
to
do
a
dash
or
one
where
we
update
the
pre-fill
tables
or
what
is
already
registered
in
the
document
and
and
not
by
later
experts
they're
suggesting
an
expert
action
and
action
so
that
that
should
be
obvious.
What
we
want
to
do
there
and
one
issue
came
up.
A
Would
it
be
useful
to
limit
the
length
of
the
names
of
these
Target
attributes,
so
Thomas
facade
asked
whether
we
maybe
want
to
go
with
a
limit
like
16
or
32.
That
can
sometimes
make
make
it
easier
for
implementations
to
handle
these
things.
A
Okay
and
once
we
have
Dash
or
one
I,
think
we
want
to
go
for
working
with
Glasgow.
B
A
Okay
and
last
slide
in
2018
I
think
I
wrote
this
Corrections
and
clarifications
draft
I
actually
had
pretty
much
empty
house.
For
that,
and
you
know
we
I
think
we
we
have
repeatedly
found
out.
We
want
to
revive
that
the
GitHub
repository
is
there.
So
if
you
want
to
look
at
recent
activity
where
recent
is
defined
in
the
last
four
years,
you
you
can
look
there
so
One
Direction.
A
Tell
us,
and
it's
also
hard
to
to
add
additional
explanation
to
a
router
reports
that
that
have
been
accepted
so
I
it's.
It
may
be
good
to
have
something
and
pull
request.
26
is
an
example
for,
for
such
a
Errata
report
being
explained
in
a
little
bit
more
more
detail,
and
we
probably
would
want
to
do
this
with
all
the
rata
that
that
are
out
there.
So
people
can
look
us
up
in
in
more
detail,
so
this
would
be
on
the
explanation
side.
A
Then
we
have
two
other
sources
of
wisdom
that
we
could
make
use
of.
One
is
the
coab
fact
frequently
ask
questions
and
there's
a
link
in
the
chat
that
contains
good
stuff.
That
currently
is
just
numbered
questions,
but
we
probably
could
sort
it
in
into
clarifications
and
explanations
and
so
on,
and
then
of
course
there
there
is
a
pretty
decent
every
Co-op
document
that
just
isn't
complete,
so
we
we
never
decided
to
finish
it.
But
what's
there
is
already
pretty
good
and
we
probably
should
think
about
what
what
to
do.
A
With
this
document.
We
just
reference
the
the
expired
internet
draft
from
Corrections
and
clarifications,
or
do
we
actually
pull
in
text
from
there
so
that
that's
what
we
need
to
decide,
I
think
I'm
going
to
make
a
pull
request
for,
for
that
shows
how
pulling
up
pulling
in
the
FAQ
might
look
like,
of
course,
the
the
FAQ
is
also
useful,
because
it
is
a
Wiki
document
and
people
can
just
contribute
to
that
with
low
effort.
A
So
we
have
to
balance
here
between
wanting
to
have
something
like
a
consensus,
working
group,
a
view
of
of
the
answer
to
the
question
which
is
really
more
going
into
the
curricula
document
and
just
having
a
random
Wiki
somewhere
on
on
the
web,
that
that
is
like
stake.
Overflow
useful
but
may
not
be
right,
and
that
brings
up
the
question
of
what
what
is
actually
going
to
be
the
intended
centers
of
this
document.
A
Is
this
a
standards
Wi-Fi
ITF?
Oh
yeah,
Wiki.
We
have
a
IDF
Wiki,
yes,
so
we
could
use
that
it.
It
has.
A
A
A
Okay,
but
but
that's
the
Spectrum
on
which
we
will
need
to
balance
things
so
I
wanted
to
bring
up
my
last
Point
here,
which
is
what
is
the
intended
status
of
this
document?
A
Of
course,
if,
if
we
ever
want
to
adopt
this
as
a
working
group
document,
then
we
should
at
least
at
that
point,
make
up
our
mind
about
that.
We
can,
of
course,
always
change
that,
so
there
are
likely
to
be
some
standard
strike
elements
in
that
documents
or
the
actual
Corrections,
which
would
need
it
to
be
based
on
a
strong
consensus
and
there
will
be
informational
Parts
in
the
document.
So
things
like
how
do
I
do
eggs
and
the
answer?
A
Maybe
don't
do
eggs,
and
this
is
certainly
not
not
standard,
striking
content
but
informational
content
and
it
might
be
a
little
bit
opinionated.
A
So
it
might
not
have
the
same
strong
consensus
that
that
the
corrections
we
are
making
have
and
yeah
from
a
process
point
of
view,
it's
nice
to
not
have
to
decide
whether
we
have
a
standard
strike
or
an
informational
resolution
at
the
time
where
we
work
on
this
item
but
yeah
from
from
a
document
point
of
view,
this
would
need
to
be
clearly
separated
right
now
there
is
a
section
two
and
a
Section
3
in
the
draft
that
have
a
similar
difference
between
them
and
yeah.
A
A
B
A
Yeah,
so
if
you
look
in
this
pull
request,
26
there
is
a
sentence
that
is
missing
there,
which
is
the
the
CT
Target
attributes
cannot
be.
Repeated,
cannot
be
used
multiple
times,
and
the
correction
is
that
that
we
Supply
this
missing
sentence,
which
the
rata
report
already
does.
But
of
course,
it's
much
better
to
have
a
true,
true
update
document
update
in
in
the
form
of
Corrections
document.
For
that,
because
the
explanation
can
be
in
more
detailed.
A
Yeah
I'm
not
so
sure
whether
these
are
clarifications
which
really
are
essentially
editorial
Corrections,
where
the
the
text
isn't
clear
enough.
While
it
is
clear
what
the
work
group
was
intending
there,
it
may
not
be
clear
to
to
a
reader
that
has
not
been
part
of
the
discussion,
so
that's
that's
different
from
a
correction
where
we
actually
something
that
that
is
missing
or
even
misstated
in
the
document.
A
B
Which
payload
and
the
payload
as
part
of
the
kashki?
Yes,
for
example,
yeah
that
that
sounds
really
standard
struck,
related
yeah.
B
A
Yeah
so,
for
instance,
we
had
a
router
report
which
needs
to
rejected
where
somebody
complained
that
we
are
not
describing
how
the
message
number
space
is
supposed
to
be
managed,
and
that
is
certainly
something
that
the
the
document
leaves
open
on
purpose
and
where
we
have
additional
sources
like
the
every
Co-op
discussion
that
can
be
used
in
coming
up
with
a
good
implementation
strategy.
A
So
this
this
would
be
an
explanation
and
even
the
recommendation,
a
weak
recommendation
of
certain
strategies,
or
maybe
an
exposition
of
strategies
that
that
have
not
run
into
opposition,
yeah
I,
think
there's
a
wide
spectrum.
There.
B
B
A
B
B
Okay,
I
I,
don't
think
you
have
a
slides
for
that,
but
it
boils
down
again
to
the
same
mail
that
Francesca
sent
a
few
weeks
ago
and
I
may
soon
share
my
screen
again.
B
Let's
see
okay
now
you
should
see
the
list
and
yeah
and
we
got
some
input
today
that
you
can
find
on
the
on
the
notes
and
then
the
mail
from
Esco
I
noticed
so
that
well
in
particular,
some
input
from
clouds
I
think.
B
B
So
we
can
go
through
them.
One
by
one
looks
like
48.95.
It
is
fine
for
close
here
and
in
general
there
were
no
objections.
There
was
the
qualification
from
Esco
I
think
yeah
to
behold
for
document
update.
Then
there
were
the
the
two
Errata
that
were
filed,
basically
together
by
by
Klaus,
about
mentioning
explicitly
the
fragment
component
of
the
URI
for
the
co-op
and
Coopers
scheme
separately,
so
that
was
initially
proposed
as
as
verified
and
and
causes
nothing
that
there
used
to
be
an
Errata
for
for
HTTP.
B
That
does
exactly
the
opposite
and
in
fact,
if
you
check
RFC
9110,
which
is
updating
that
I
see
that
rather
referred
to
yeah,
you
don't
have
the
fragment
component
mentioned,
and
the
mail
from
Esco
today
will
further
build
on
that.
If
I
understand,
let's
go
correctly,
saying
that
they'll
introduce
confusion,
basically
introducing
the
fragment
component.
We
introduce
confusion
in
in
the
light
of
this
comments.
B
That
yeah
it
makes
sense
to
me
would
be
the
best
thing
to
do
basically
rejecting
the
Errata
that,
by
the
way
calls
himself
proposed
yes,
but.
A
I
remember
what
happened
so
Klaus
found
this
this.
There
is
a
problem
there
and
Klaus
found
it
and
wrote
it
up,
but
then
we
learned
that
that
HTTP
actually
had
the
same
problem
and
people
fixed
it
in
the
other
direction
and
we
probably
shouldn't
be
doing
something
that
is
different
from
HTTP
for
yeah.
You
may
not
like
the
way
this
has
been
solved
in
HTTP,
but
it's
kind
of
the
the
received
interpretation
of
your
eyes
so.
B
So
last
call
is
on
Francesca
I
guess,
but
this
is
pointing
to
rejecting
the
rat
actually
49
46
and
49
47.
B
B
A
Yeah,
so
really
this
should
be
hold
for
document
update,
because
it's
one
of
those
places
where
we
have
to
have
more
discussion
to
actually
correctly
address
the
report
and
the
the
distort
point
with
the
document
where
we
do
this.
B
Okay,
then
49
48
looks
good
too
close
to.
He
has
a
note,
though,
that
this
is
a
substantial
protocol
change,
yes,
that
that's
what
we
were
discussing
before
and
I
believe
it
can
be
one
point
that
warrants
the
standard
struck
status
to
core
class,
for
example,
but.
A
So
again,
it's
a
clarification
for
me,
because
that
was
always
what
we
intended,
but
yeah.
The
text
doesn't
currently
reveal
that
this
is
what
we
intended.
B
B
Okay,
so
we
we
take
this
as
verified,
I
think
49
49
was
fine,
49
54.
B
This
was
about
oh
right,
The
Columns
of
the
Registries
and
yeah
clouds
on
again
on
its
own
raised
the
rata
as
as
a
comment
and
yeah
I
think
this
latest
comment
is
correct.
If
you
want
to
express
both
the
media
type
and
the
parameter
in
the
content
type
column,
so.
A
B
Yeah
but
basically,
what's
the
outcome
about
handling
this
Errata
should
it
be
rejected
because
it
needs
to
be
edited
at
the
very
least
right
to
to
be
verified.
B
So
it
can
be
rejected
and
we
know
that
this
will
be
fixed
in
complex,
for
example,
or
edited
and
then
hold
for
dock
update.
B
B
So
this
would
be
about
editing
the
current
rata
and
then
take
it
as
all
for
document
update.
C
C
B
A
C
B
Okay,
but
one
step
at
a
time,
so
50
78
should
be
good
and
then
we
should
double
check
again
instead
52
54.,
so
that's
about
66.90
actually
link
format,
and
it
was
a
proposal
that
I
do
remember
in
detail,
but
it
was
about
yeah,
updating
the
the
format
and
and
the
semantics
of
links
to
rule
out
a
variant
that
was
very
error
prone
with
strange
Uris,
a
yeah.
B
B
So
I
I
can
agree,
it
is
a
concern
and
the
fix
looks
like
it
works,
but
isn't
it
actually
changing
the
intent
and
the
grid
intent
of
66.90?
This
looks
something
to
be
fixed
with
an
update
to
66.90
right.
A
Yeah,
so
we
would
invalidate
all
those
cases
where
people
have
used
an
HTTP
URI.
That
is
innocuous
from
the
point
of
view
of
the
link
common
syntax
without
codes.
A
B
A
So
that
that's
the
the
the
real
problem
is
is
much
deeper.
The
real
problem
is
that
the
the
a
b
and
F
tells
you
the
syntax
of
of
The
Interchange
thing,
but
it
doesn't
tell
you
what
meaning
you
actually
derive
from
from
that
and
it's
the
meaning
that
you
derive
from
that.
That
then
again
requires
a
syntax
and
that's
what
8288
did
so.
The
syntax
they
have
is
is
the
ABN
F
Productions.
They
have
are
mostly
not
about
the
interchanged
data,
but
about
what
the
the
decoding
process
of
The
Interchange
data
derives.
A
And
then
you
have
ABN
F.
That
tells
you
what
that
can
be,
and
we
we
would
kind
of
move
away
from
this
again
by
saying
even
the
simplest
HTTP
Uris
need
to
be
quoted,
and
this
would
move
quoting
again
from
something
you
you
just
do,
because
this
index
otherwise
wouldn't
work
lexically
to
something
that
has
weird
semantic
consequences.
A
Well,
6690
is
kind
of
inconsistent
because
it
does
rewire
quoted
Syntax
for
for
a
couple
cases,
which
is
not
the
right
thing
to
do.
But
it's
it's
exactly
this
dual
dual
layer,
ABN
F
issue
that
we
have
in
so
many
other
places.
A
Yeah
so,
for
instance,
the
anchor
always
has
to
use
quotes.
B
B
So
I
think
the
conclusion
was
to
do
a
rejection
and
then
let
corclair
do
the
work.
But
the
note
on
Carson
says
and
helpful
document
update.
Instead.
C
B
C
A
B
Clarifications
will
come
in
corclair
anyway:
okay,
good,
so
just
double
checking
all
is
clear
for
the
controversial
ones.
So
this
one
this
one
edited
them
verified
and
this
one
yeah
yeah.
B
Okay,
we
have
one
more
to
check
5284.
B
If
I
understand,
Klaus
comment
correctly
is
arguing
that
you
don't
need
to
restate
that
the
token
is
already
per
request
given
given
a
pair
of
endpoints.
So
it's
a
not
useful
repetition.
A
B
A
With
the
term
unique-
and
it
doesn't
get
better
by
adding
per
request
to
to
that
sentence,
so
I
I
probably
can
guess
what
the
sentence
is
trying
to
say
without
that
Edition.
But
with
that
Edition
I'm
I'm
completely.
Let.
A
B
Francesca
is
wondering
for
Klaus
may
there's
no
mail
from
Klaus
Francesca.
This
table
just
appeared
a
few
hours
ago
in
the
notes
and
well
it's
pretty
comprehensive.
I
I
think
it
was
good
material,
but
no
mail.
B
B
But
Carson,
if
you
understand
your
point
anyway,
the
addition
from
Med
is
not
really
clarifying
anything.
B
A
B
A
B
Okay,
okay,
then
5284.
If
it's
fine
to
Francesca
can
be
rejected
too.
B
Okay,
Francesca
left
anyway,
foreign
I'll
try
to
build
up
in
these
minutes
a
summary
for
Francesca
to
consider
as
final
input
from
from
the
group.
But
those
were
clouds
said:
okay
are
confirmed
basically
and
then
in
these
detail
notes
we
took
decision
on
on
the
other
ones.
B
A
That
was
about
sending
an
error
response
to
a
response.
That's
not
possible.
B
Cool
okay:
if
there
are
another
comments
on
the
Arata
resolution
we
have
covered
this.
B
B
Well,
you
know
it
comes
soon
when
it
comes
enjoy
the
Christmas
season.