►
From YouTube: IETF-CELLAR-20220125-2000
Description
CELLAR meeting session at IETF
2022/01/25 2000
https://datatracker.ietf.org/meeting//proceedings/
A
B
B
B
C
B
So
you'll
welcome
to
everybody
as
you're
joining
in
I'm
just
going
to
put
up
some
info
here.
So
I'm
not
sure
if
you
can
see
my
cursor.
Yes,
you
can.
So
this
is
the
mute
button.
This
turns
on
your
video.
You
can
do
that
anytime,
you
like,
if
you
want
to
interrupt
or
interject
and
since
we're
quite
a
small
group
that
should
be
not
a
problem.
This
is
the
request
to
present
that
I
have
to
approve.
B
This
is
used
to
present
slides
that
are
already
uploading.
We
don't
use
that
much
and
this
is,
to
put
your
hand
up
or
join
the
queue.
If
we
have
a
list
of
people
that
are
waiting
to
go
and
we're
not
such
a
small
group,
we
probably
won't
do
that.
So
I'm
just
going
to
also
point
at
the
mute
button
there.
It
is.
B
Try
unmuting
and
in
some
browsers
you
have
to
give
mic
permission
each
time
which
is
dumb
and
others
you
don't
so
anyone
else
want
to
unmute
and
say
something.
B
I
I
thought
it
was
a
good
hello.
One
problem
I
have
with
me
deco
is
that
mute
on
a
mute
is
actually
has
like
a
fair
bit
of
latency
in
the
sense
that
you
click
on
mute.
You
say
hello,
and
then
you
and
you
mute
again,
and
usually
it
didn't
actually
happen,
so
you
actually
have
to
wait
for
it
to
finish
this
little
animation
when
it
unmutes
you.
B
B
B
Okay,
spencer:
has
a
spencer's
boss
enforced
a
a
forced,
a
meeting
on
him
that
ends
in
about
15
minutes,
so
he'll
be
joining
us.
Then.
B
So,
while
you're
sitting
here
waiting
for
another
minute,
you
could
go
to
the
the
draft
agenda
and
scroll
to
the
bottom,
or
I
could
do
that
and
we
could
look
at
the
minutes
from
the
previous
meeting.
If
you
were
there,
you
have
any
comments
about
that
and
in
the
chat
I
will
paste
the
url
as
well.
E
B
Hear
you
anymore,
I
usually
use
this
one
online
mic
test
to
find
out
whether
or
not
it
works
and
apparently
it
all
lives
in
your
browser.
It
doesn't
actually
announce
anything.
So
that's
I
like
that
one
anyway
to
know
for
sure.
If
it's
it's
really
getting
all
the
way
through.
B
All
right,
so
I
guess
we
should
start
here.
So
I'm
going
to
got
some
names
in
here.
Wonderful,
martin,
steve
michael
meter
here
and
that's
it
there's
four
of
us.
We
have
christian
who's,
just
a
lurker
by
jabber
and
jerome.
B
Okay,
so
we've
all
seen
the
note
well
many
times
there
it
is
and
the
major
one
is
if
you're
aware
of
any
patent
or
patent
applications,
it's
controlled
by
you
or
your
sponsor,
you
must
disclose
it,
and
you
agree
that
this
session
may
be
recorded.
B
And
if
you
don't
want
your
face
shown
then
don't
turn
on
your
video,
and
we
also
have
this
code
of
conduct
that
we're
trying
to
make
sure
that
everyone
is
aware
of
it's
been
around
for
quite
some
time.
But
people
want
to
make
sure
that
people
really
seen
it
it's
contained
in
bcp,
25
and
54.
B
So
this
is
the
status
of
our
drafts
today.
The
only
real
update
is
ffv1
was
reposted
a
couple
days
after
the
last
meeting.
Thank
you
very
much
and
the
rest
of
them.
Probably
we
could
be.
You
know
posting
progress
more
particularly
matraska
as
we
go
along,
so
previous
status
was
talked
about
jerome
and
there's
this
action
here
that
I've
kept
for
old
business.
So
we're
going
to
come
back
to
that
when
spencer's
around.
B
We
had
the
question
before
which
we
did
not
answer,
whether
we're
going
to
do
an
update
for
ebml
and
whether
we're
ready
to
actually
submit
a
working
group
draft
yet
to
include
whatever
errata
that
has
arrived,
and
you
know
just
because
we
post
it
doesn't
mean
it's
done.
It's
just
there,
but
we
did
agree
before
that.
It
was
after
matraska
is
out
and
we're
not
going
to
proceed
until
matraska
is
out.
So
we
know
that
if
there's
any
errata
that
we
can
include
it.
B
Yeah
also
one
of
the
things
that
we
can
do
if
we
have
some
imp,
two
implementations.
B
So
when
was
this
came
out,
this
came
out
in
2020
in
in
in
july,
so
and
its
status
is
that
it
is
a
proposed
standard
and
so
the
the
ietf
has
now
two
levels
of
standards.
We
have
proposed
standard
and
we
have
internet
standard.
We
used
to
have
three
levels
proposed
draft
and
full,
and
we
collapsed
the
draft
and
the
full
into
one
one
thing,
and
so
after
document
has
had,
has
been
out
there
for
two
years
and
has
at
least
two
interoperable
implementations.
B
Then
we
can
ask
the
document
to
be
revised
to
be
internet
standard
and
one
way
is
that
it
just
it's
just
if
there's
no
text
changes
required,
then
it's
just
an
administrative
bit.
That's
twit
tweaked!
Since
we
are
changing
the
document
and
proving
it
then
we
could
we
could
if
we
thought
it
was
ready
to
make
that
propose
that
as
a
as
a
to
be
an
internet
standard,
and
it's
just
it's
like
getting
an
extra
medal
at
the
end
of
the
race
really.
B
No,
if
we
had
further
rada,
that
may
argue
for
it
could
argue
that
we
weren't
really
ready,
or
it
could
argue
that
we
just
have
to.
We
need
to
accept
it.
So
I
think
that
that
I
think
that
that
decision,
I
think
we
should
leave
to
the
very
end.
A
By
different
implementations,
do
they
have
to
all
do
all
the
what's
in
the
document?
For
example,
there's
the
you
know
the
unknown
size,
that's
not
well
supported
in
many
players.
B
Well,
so
there's
there's,
there's
two
things:
one
is
that
if
it
turns
out
there's
bits
and
pieces
that
are
not
completely
well
implemented,
then
we
may
actually
note
that
in
the
document
and
say
these
are
not
have
not
yet
occurred.
A
second
possibility
is
that
we
would
say
they
are
not
implemented
and
no
one
cares
about
them
and
we
should
remove
them,
and
a
third
possibility
is
that
we
should
wait
and
not
declare
it
as
an
internet
standard
until
we
have
seen
a
more
significant
amount
of
interoperability.
B
So
you
know
if
we
were
building
a
vpn
product
and
it
turned
out
that
no
one
had
actually
tried
doing
encryption,
then
you
might
say
that.
Well,
it's
not
really.
It's
not
really
a
standard
yet
because
no
one's
actually
completed
the
private
part,
just
the
virtual
network
part
so
that
that
might
argue
for
okay,
we're
not
ready.
But
if
someone
said
oh
well,
we
never
actually
tried
encrypting
with
256-bit
keys.
B
Well,
we
might
say:
well,
that's
just
a
detail
and
we
know
it
works
with
128-bit
keys.
So
that's
good
enough
right
later
later,
people
when
they
feel
the
need
for
stronger
encryption
will
do
that
and
that's
okay.
A
And
I
think
we're
in
a
different
position
from
all
of
that
is
that
it's
quite
an
old
format
now
actually
and
there's
lots
of
implementations.
A
B
I
agree
with
you,
I
mean
exactly
so
I
don't
think
that's
a
that's
a
that's
an
issue
that
in
fact,
may
argue
for
that.
The
subset
that
that
those
old
implementations
implement
should
be
noted
that
if
you
want
to
be
fully
interoperable
with
everything,
then
don't
use
this
feature
this
feature,
or
this
feature
right
and
that's
okay.
They
can
still
exist
in
the
document.
It's
just.
They
are
not
fully.
You
know
standardized
yet
or
there's
some
caveats
if
you
use
them
or
you
know
something
like
that
to
that.
B
So
now-
and
the
other
question
about
this
is
to
what
extent
is
this
about
ebml
versus?
Is
it
about
matraska,
and
it
might
be
that
we
actually
can't
advance
until
matraska
advances
because
you
actually
have
to,
and
you
actually
have
to
have
something
to
play
right,
so
I
suspect
I
suspect
that
may
be
the
case.
B
Okay,
so
with
that,
let's
go
on
to
matraska
on
the
whatever
issues
that
we
think
we
need
to
discuss
as
a
group
there
are
and
that
kind
of
stuff.
Let
me
see
you
found
the
mute
buttons
I'll
close,
that.
A
I
I
merged
a
few
ones
recently,
because
I
think
there
was
consensus,
also
updated
part
of
rewrite
of
the
time
definitions,
because
there
are
two
or
three
different
time
bases
in
matroska
and
before
it
was
all
scattered
in
the
document.
Now
I
just
define
different
ticks
for
each
clock
or
in
each
time
base.
A
A
A
A
Something
that
martin
bellow
was
talking
about
a
lot
in
the
past,
but
chapter
time
end
being
mandatory
or
not,
and
I
found
that
they,
it
was
needed
to
be
mandatory
for
other
chapters,
but
then
in
the
example
I
made,
I
realized
that
if
you
go
one
level
up,
it
just
doesn't
work
anymore.
A
So
now
only
the
lower
level
of
nested
chapters
in
men
is
mandatory.
The
others
are
not
and
actually
should
be
discarded.
A
So
I
changed
that
text
again
and
I
think
we
need
some
more
review
on
that
there's
also
578.
A
B
Or
something
no,
they
they
would
like.
The
summary
in
the
ayana
considerations,
area,
yeah,
okay,.
A
A
I
saw
on
ebml
that
for
every
value,
whenever
there's
a
gap,
they
add
an
entry
and
say
well.
This
is
unused
for
now
unassigned,
because
in
xslt
I
don't
know
how
to
do
that.
I
didn't
do
it,
so
they
would
still
need
to
generate
that
part,
but
at
least
the
actual
values
they
are
all
there
sorted
the
way
they.
A
B
I
think
that's
fine,
I
so
the
rfc
editor
might
reformat
that
in
the
final
document
or
oh.
A
A
A
A
A
That's
great,
but
why
yeah
one
other
thing
compared
to
what
they
have
is
that
in
their
case
the
each
element
is
linked
to
the
section
where
it's
defined
with
the
number
of
the
section.
But
okay,
we
don't
know
that
number
until
the
fire
is
generated,
and
then
it's
separated
generally
and
separately,
and
we
have
no
idea
actually
to
know
which
of
the
anchors
we
have
in
the
source
goes
to
what
value
in
the
output.
So
I
didn't
put
the
actual
numbers.
I
put
the
anchors
that
we
use.
B
Know
I
I
think
we
could
fix
that
up
afterwards.
I
don't.
I
think
that
it's
unusual
for
people
to
generate
this
mechanically
like
this,
but
we
could
generate
it
mechanically
once
and
when
we're
done
and
then
we
could
fix
it
if
we
need
to
yeah.
B
So,
are
you
ready
to
post
a
new
version
of
the
document.
A
I
would
like
5
21
to
be
merged.
I
hope
this
week
it
can
be
done.
It's
the
microscopics,
so
basically
the
two
or
three
time
bases.
I
think
it's
okay,
the
fonts.
I
I
don't
remember
what
this
it
says.
There's
a
change
requested
from
moritz,
but
I
think
it's
actually
okay.
We,
I
did
what
everything
he
asked.
A
B
So
let
me
let
me
ask
this
question,
so
we
have
things
marked
bug
in
the
issues
and
in
the
pull
requests,
and
so
I.
G
B
B
Okay,
so
what
I'm
trying
to
get
at
is
that,
is
there
a?
Is
there
a?
How
are
things
that
issues
or
pull
requests
that
you
feel
need
to
be
completed
for
the
document
to
be
complete?
C
A
Yeah,
so
I
use
that
link
which
is
basically
filtering
only
the
matroska
specifications
so
leaving
out
the
codec
things,
the
tags
and
also
some
label
control
track,
which
is
some
other
document
as
well,
also
mature
scav5,
which
is
basically
everything
we'll
add
in
the
future,
even
though
it
could
be
done
now.
So
basically,
that
just
gives
you
exactly
the
list
of
what's
left,
in
my
opinion,
for
the
document,
and
so
if
you
look
on
the
right,
almost
all
of
them
has
a
linked
full
request.
A
So
that's
basically
the
only
pull
request
we
need
to
fix
after
that
will
be
good,
the
ones
that
don't
have
a
pull
request.
I
don't
remember.
A
A
A
Oh,
yes,
I
don't
think
it's
a
priority,
or
even
something
that
will
end
up
in
the
specs.
It
was
just
a
way
to
describe
the
metro
scale,
but
in
text
right
yeah
it
doesn't
fall
in
the
actual
spec,
so
yeah
they
all
have
a
linked
request
that
needs
to
be
merged
and
so
basically
the
other
link.
I
use.
A
Which
went
through
is
the
same
filter,
but
for
pull
requests,
and
in
that
list
you
have
nine
pull
requests
open,
which
more
or
less
correspond
to
the
issues
on
the
other
tab:
okay,
cool!
So
once
we
have
closed
these
nine.
For
me,
the
document
is
ready
to
be
published.
B
Cool
all
right,
that's
wonderful!
Do
you
want
to
publish?
Is
there
a
point
in
the
next
two
months
that
you
think
you
want
to
publish?
You
want
to
publish
an
inter
on
a
new
version,
or
do
we
need
to
wait
for
all
nine
to
be
done.
A
B
A
B
A
B
So
jerome,
I
would
love
to
have
the
same
conversation
about
flack.
Oh
no,
sorry,
martin
is
working
on
flak.
Do
you
want
me
to
pull
that
up?
Where
is
it.
B
B
F
Well,
the
the
thing
I
mentioned
in
the
email-
I'm
not
really
sure
whether
some
of
the
pull
requests
add
something
to
the
document.
That
is
what's
the
word
appropriate
for
for
internet
standards,
okay,.
B
F
I
I
added
two
appendices,
one
with
numerical
considerations
and
one
with
well
really
for
both
examples.
F
F
Well
the
examples.
Mostly
they
help
with
that
when
you,
when
for
people
well,
the
the
document
is
quite
how
do
you
say
it.
B
Yeah,
no,
I
agree
I
I
actually.
This
is
very
good
kind
of
appendix
kind
of
content
it.
Actually,
if
it's
helpful
to
an
implementer,
then
I
think
that
that
concluded
we
we
do.
We
don't
always
do
this
in
itf
documents,
it's
probably
a
bug
that
we
don't
do
it
more
often,
but
implementation
advice,
particularly
in
a
non-normatively
in
an
appendix
makes
sense
to
me.
F
Well,
that's
the
numerical
consideration
section
this
and
this
in
well.
This
section
this
section
with
the
examples
are
three
example.
Example
files
with
an
well
files
files.
You
can
there's
a
hexadecimal
interpretation.
You
can
really
convert
into
a
flag
file
which
works
with
coder.
That's
older,
that's
compliant
to
the
format,
and
then,
in
this
section
it
really
says
at
this
point
at
this
byte.
You
see
so
many
bits
that
code
for
this.
D
No
yeah,
this
part,
I
was
thinking
to
the
other,
pier
appendix
numeric
numerical
considerations.
This
form
it
took
like
more
some
definitions
and
some
vocabulary
about
the
audio,
and
it
may
be
good
to
put
it
in
introduction.
Maybe.
B
D
Of
consideration
now
this
is
the
numerical
considerations,
but
it
looks
like
that.
Some
part
of
that
is
used
is
useful
for
understanding
the
specification
itself.
After
am
I
wrong
or
yeah.
F
Well,
well,
I
don't
feel
that
because,
because
well,
when
you
choose
a
64-bit
sign
data
type
for
every
all
arithmetic
in
any
flock
decoder,
encoder,
you're,
safe
and
that
it
works
the
thing
with
the
convertible
consideration
is
that
usually
using
a
60
44
bit
signed
data
type,
isn't
really
the
fastest
way
to
implement
a
decoder.
F
A
D
F
Faster
you
you
have
to
or
when
you
have
to
implement
it
in
a
dsp,
dsp
or
something
a
portable
device.
You
cannot
always
use
60
for
64-bit
design
data.
You
can.
You
can
use
decision
to
see.
When
can
I
use
32
bits
or
16
bit
even
with
the
decoder
decoder
and
do
when
you
have
to
create
multiple
data
paths,
one
for
a
wide
and
one
for
a
small
small
dip?
B
D
Okay
and
for
the
example
yeah,
it
is
relatively
big
and
I
am
wondering
if
it
is
useful
to
be
in
the
afc
or
in
a
document
in
the
flac
repository.
D
F
I
don't
know,
perhaps
perhaps
also
has
elsewhere
but
yeah
but
sure
but
sure
what
kind
of
torii
story
is.
B
Extremely
valuable
in
the
document,
these
are
worked
out
examples,
so
that's
really
kind
of
really
good,
because
it
means
that
someone
can
come
back
and
feed
that
through
their
implementation
and
see
that
they
get
the
values
here.
This
is
like
awesome
that
you
know
that
it
says
each
one
there's
four
bits
and
what
the
value
is
and
what
does
it
mean.
D
B
So
basically,
what
you
say
you
say
this
informational
in
for
emotional
appendix
that's,
basically
it.
The
other
thing
I
recommend
is
that
you
start
all
of
your
sentences
on
a
new
line,
but
I'm
not
going
to
change
that
for
that.
I'm
not
there
that
just
it
just
makes
it
easier
to
edit
later
on,
okay,
okay,
but
it's
best
to
do
that
as
a
single
commit
that
changes.
The
white
space
essentially.
F
Yes,
it
obviously
has
to
project
very
well.
That's
why
I
do
a
tenth
of
each
end
of
each
example.
I
well.
I
did
everything,
obviously,
obviously,
so
I
did
check
with
the
md
signature
juror
and
that
and
that
that
failure
correct
correct,
but
there
can
can
be
a
mistake
in
there.
B
Sorry
I
was
being
interrupted,
so
I
heard
I
heard
steve
say
that
the
example
has
to
be
good
and
can't
lead
people
into
the
wrong
direction,
and
I
agree
with
that.
So
that
just
really
means
that
someone
else
has
to
walk
through
who
understands
who
probably
he's
implemented
as
well.
That's
really,
I
think,
the
needed
part,
and
as
long
as
we
can
do
that,
then
I
think
we're
good.
F
One
of
the
when
I
was
writing
the
examples
I
actually
found
a
few
bits
in
the
specification
that
well
needed
a
little
more
clarification.
Yes,
just
before
this
meeting,
I
actually
I
found
some
that
that
I
well
example.
B
Okay,
cool,
so
I
I
think
you're
making
great
progress.
I
think
you
should
merge
your
things
and
post
a
new
version
as
soon
as
you're
comfortable
with
that,
and
then
you
know
we
can
come
back
and
figure
out
what
issues
are
are
in
this
list
that
we
need
to
address,
or
which
ones
are.
B
B
All
right
in
the
name
of
time,
let's
go
on
to
talk
about
ffv1.
If
there's
something
to
talk
about.
B
B
F
F
B
F
Does
that
mean
two
matroska
implementation
should
be
fine,
or
should
it
be
two
implementations
of
ebml
one
of
matraska
and
some
of
someone
else
using
for
something
for
something
different
than
matroska.
B
It
actually,
I
think,
is
better
if
it's
to
matraska,
because
if
it
was
something
who
implemented
bob
with
ebml
and
someone
that
implemented
frank
with
ebml,
we
wouldn't
expect
them
to
interoperate
at
all.
Would
we
right
because
they
don't
do
the.
B
A
B
B
All
right
so
I'll
talk
to
you
guys
on
the
22nd
stay
warm
if
it's
cold
there,
because
it
is
here
holy
it's
cold
here
and
that's
about
it.