►
From YouTube: IETF106-AVTCORE-20191122-1220
Description
AVTCORE meeting session at IETF106
2019/11/22 1220
https://datatracker.ietf.org/meeting/106/proceedings/
B
A
B
A
D
A
A
A
A
Agenda
so
rachel
is
not
here.
She
is
switching
jobs,
switching
roles
at
her
current
at
Huawei,
so
it's
not
clear,
she'll
be
able
to
be
coming
to
the
idea
or
I.
Think
it
is
actually
is
clear
that
she
will
not
be
able
to
be
coming
to
the
IETF
anymore.
So
probably
want
to
talk
to
Barry
about
that
at
some
point.
If
he
still
thinks
we
need
three
chairs
so.
E
A
A
B
So
the
opening
is
one
of
the
here
I'm
speaking
as
one
of
the
authors
of
the
document
editor,
we
missed
during
the
process
to
an
address,
Bernard's
comment,
and
hopefully
I
will
do
it
next
week.
So
I'll
probably
have
to
ask
probably
from
Bernard
for
some
clarification
but
I'll
try
to
do
it
next
week
and
the
last
one
di
feedback,
CC
Fitbit
message:
that's
the
one
that
Colleen
is
going
to
present
when
he
comes
to.
F
Most
nineties
editor
of
the
frame
working,
so
we
updated
to
version
10
based
on
Dale's
substantial
feedback.
Let
me
preface
this
with
it
was.
It
was
quite
a
bit
of
quite
a
quite
a
few
comments
and
I
think
several
of
them
required
some
substantive
changes,
so
unfortunate
I
think
we
do
need
to
go
through
another
working
with
last
call
to
make
sure
that
those
changes
are
ok
with
everyone.
I'll
summarize
them
briefly,
but
I
think
people
are
gonna,
need
some
more
time
to
digest
it
and
make
sure
that
we're
not
breaking
anything
with
these
changes.
F
So
the
the
first
major
item
was
and
I
guess
we're
on
a
you
should
pay
attention
to
this.
The
floor
reserved
bits
that
used
to
be
in
the
short
format.
You
know
where
the
with
a
temple
area,
ID
and
the
and
the
base
layer
SiC
bit
with
those
bottom
four
bits
were
all
zeros
in
the
short
extension
format,
and
since
we
added
the
ability
to
omit
in
the
long
format,
the
other
parameters,
so
that
made
it
ambiguous
whether
something
was
dead,
zero
with
B
equals
zero
or
whether
something
was
a
non
scalable.
F
Extension,
where
you
should
ignore
those,
so
it's
really
not
possible
to
reserve
them
for
future
use,
which
one
is
draft,
was
relying
on
them
being
available
for
future
use
for
the
non
scheduled
extensions.
Only
if
you
knew
that
beforehand
that
you're
only
ever
going
to
use
non-scalable
streams
through
some
other
signaling,
you
knew
that
there
was
only
non-scalable
screws.
That's
not
time.
You
can
really
reuse
those
bits.
If
you
had
a
case
where
there
could
be
scalable
and
on
scaleable
streams,
then
there's
no
way
you
could
reuse
those
bits
because
there's
no
distinguish
the
two.
A
F
I
mean
this
thing
is
just
to
have
a
different.
You
are
n
or
n
parameter
after
three
marking,
but
that's
essentially
a
different
extension
ID,
so
you
could
differentiate
them
that
way,
but
that
essentially
is
a
different
header
extension
at
that
point.
So
if
you
know,
if
Ronnie
wanted
to
pursue
the
priority
work,
you
just
have
a
different.
F
So
everything
only
depends
on
the
base
layer
and
they'll
pointed
out
that
it
was
an
oddity
that
the
baby
that
would
always
need
to
be
one
in
that
in
that
case,
because
you
you
only
on
the
base
layer,
but
in
other
cases
we
say
that
the
be
a
bit
of
zero,
for
example,
for
non
scribble
streams
so
to
clean
that
up,
I
just
forced
the
nib.
It
always
be
zero
when
the
ten
is
zero.
That
way,
it's
clear
that
the
natural
the
natural
case
is
when
something
is
ignored
at
zero.
F
So
then
you
don't
have
to
always
set
it
to
1
and
then
in
the
base
layer.
It's
just
always
zero
in
the
base
layer.
So
that
seems
a
little
bit
cleaner.
That's
a
substantive
change
and
the
next
changes
tells
your
epic
index.
The
current
draft,
the
older
draft
said
that
it
counts
the
typical
zero
frames
and
that's
not
completely
accurate.
That's
from
the
combination
of
typical
zero
and
lid
equals
zero.
F
Next
change
is
that
if
you
don't
have
a
tid,
it's
implicitly
zero,
so
basically
non-scalable
streams
are
flat
and
they
they
only
have
that
simple
base.
Tipler's
ero
and
similarly
lids
are
implicitly
zero
when
you're
using
the
a
short
extension.
So
if
you
don't
specify
it
you're
in
the
base
special
layer
and
if
you
omit
it
in
the
long
extension
same
thing,
it's
implicitly
zero
some
text
to
clarify
that
that's
not
stuffs
into
this
just
clarification:
they
are
ID
mappings.
That's
also
just
clarification.
F
D
F
Obvious
to
some
other
readers
that,
if
you're
having
non-scalable
stream,
there's
no
layer,
IDs
and
so
there's
no
mappings
clarified,
that
must
rest
and
then
finally,
the
editorial
changed
just
to
not
show
the
header
extension
ID
with
a
specific
value.
It
was
more
the
specific
value
in
some
of
the
sections
and
it
should
be
a
question
mark.
That's
negotiated
by
signaling.
That's
all
changes,
but
I
think
we
do
have
to
go
through
the
work
with
boss
called
heaven.
Everybody
digest
those.
My.
G
Bernardo
bulla
Microsoft
I
have
just
a
clarifying
question
for
you
mo
because
the
fray
marking
draft
is
relatively
simple
to
implement
it's
attractive
to
play
with
for
things
like
a
v1,
and
the
question
has
arisen
as
to
which
loads
of
baby
one
it
is
applicable.
For
so
my
taking
tell
me
I'm
wrong
or
not
is
that
it
applies
for
the
tempo
modes
and
this
normal
kind
of
spatial
modes,
but
not
for
the
KS
v
c
or
c
shift
modes
or
any
of
those
weird
things
or
this.
F
So
if,
if
people
wanted
to
go
outside
of
what
81
specifies
and
bind
to
this
specification,
yeah
the
cases
where
you
haven't
you
know,
nested
temple
hierarchies
would
would
be
usable
by
this.
But
that's
kind
of
is
difficult
to
say
other
and
that
we
should
recommend
and
have
a
section
481
about
this,
because
you
know
the
first
of
all.
There
is
no
any
one
art
speed
you
have
not
in
not
ITF
draft
for
that
and
if
there
ever
were
one
it's
my
understanding
is
it's
likely
not
to
specify
a
binding
to
frame
marking.
G
Mention
this
because
a
frame
marking
is
currently
in
library,
RTC,
and
so
when
a
v1
is
added
in.
You
have
to
figure
out
what
do
you
do
with
it?
And
even
if
we
don't
mention
a
v1,
just
understanding
what
kinds
of
modes,
because
some
of
these
modes
also
exist
in
vp9,
just
understanding
the
applicability
of.
H
B
Finally,
thank
you,
so
just
it
just
for
clarification
is,
as
far
as
I
understand
that
for
the
frame
out
theme
for
any
new
codecs,
which
are
not
part
of
this
document,
didn't
need
to
be
specified
elsewhere,
because
we
are
closing
this
one.
So
if
you
don't
need
for
everyone,
you
need
someplace
to
write.
It.
A
Yeah
I
mean
I
think
the
vp9
draft
does
mention
its
frame,
marking
mappings
in
it.
But
again
it's
only
I
think
frame
marking
is
pretty
clear
that
it's
only
for
these
simple
and
common
scalability
structures-
and
you
know
all
codecs
can
do
weird
things
only,
but
you
know
only
difference
is
how
easily
can
they
describe
them,
but
they
can
all
be
weird
and
frame.
Working
is
only
for
the
things
we
thought
were
the
simple
cases
when
we
were
writing
it.
G
H
G
F
Mean
you
know,
given
that
those
are
just
you
know
just
recently
within
a
few
months.
You
know
I
mentioned,
and
you
know
widespread
deployed,
so
I
I
belittle
averse
to
starting
to
enumerate
things
that
are
newish
because
that
list
could
be.
You
know,
I
worked
really
long
and
if
they
never
see
deployment,
then
it's
kind
of
you
know.
F
We
mentioned
those
things,
I
think
it's
cleaner
and
what
want
any
might
things
and
then
have
someone
say?
Oh
well,
you
didn't
write
this,
so
it
must
be
in
scope.
So
it's
kind
of
ambiguous
whether
or
not
something
that's
not
her
enumerated
you
know
is
in
or
out
of
scope,
so
I
think
it's
cleaner.
Just
to
say
only
these
very,
very
well-defined
streams
are
in
scope
and
everything
else
is
out
of
scope.
B
Okay,
big:
what
will
happen
so
that's
why
we
have
either
be
so.
We
make
a
new
definition
for
frame
mounting.
Like
you
suggested,
we
may
need
a
different,
a
different
structure,
a
different,
a
different
bill,
other
expense
extension
for
a
first
specific
one.
So,
but
that's
we
don't
have
to
decide
it
now,
because
now
we
are
closing
based
on
the
current
text,
that
there
is
a
current
called
exit
out.
There.
E
G
A
G
F
A
F
Mean
it's
a
the
creator
of
the
structure
knows
better
than
we
do
whether
or
not
it
would
work
with
frame
markings
so
I
think
to
actually
prohibit
it,
and
you
would
be
the
wrong
thing.
You
know
if
I'm,
if
I'm,
creating
a
complex
structure
but
I'm
clever
enough
to
make
it
work
the
same
as
nested
structures,
then
I
should
be
prohibited
from
being
able
to
use
the
frame
marking.
You
know
approach,
but
I
could
go
either
way
and
I
prefer
to
leave
the
text
as
is
not
recommended,
but
I
was
a
little
too
strong.
A
A
B
A
B
B
E
B
I'm,
not
sure
I,
think
I
think
they
that's
I'll
have
to
look
it
because
my
understanding
is
that
there
was
the
issue
about
how
it's
related
to
to
internally
into
the
text.
That's
in
the
TSA
Vietnamese
point,
for
that,
is
it
a
normative
reference
information?
Informative?
What
to
do
is
it
or
to
include
the
text?
I!
Think
that's
the
point
that
is
open
there
right.
E
B
B
A
Yeah,
so
I
did
a
minor,
refresh
and
I
think
it
was
a
one
minor
update,
but
there's
still
it's
still
basically
waiting
on
a
bunch
of
informative
texts,
which
I
was
hoping
that
my
co-authors
would
write
and
they
haven't.
So
it
might
be
that
we
need
to
find
some
other
volunteer
to
write
that
text
just
sort
of
like
an
overview
of
how
the
vp9
codec
as
a
whole
works.
I
mean
this
has
been.
A
J
A
B
I
B
B
B
We
have
to
go
back
to
the
goals
in
milestone
and
you
think
see
that
what
to
need
update,
but
currently
we
have
the
tetrarch's
already
passed
you,
the
RTP
header
extension
for
video
frame.
Hopefully
we
probably
hopefully
will
have
it
by
January
or
February,
we'll
be
able
to
send
it
for
publication.
So
we
need
to
update
this
this
date,
the
that's
the
June
2019,
that's
the
draft
for
calling,
hopefully
is
done.
B
We
hear
about
it
when
he
does
this
presentation,
because
we
will
ask
why
it's
not
finished
yet,
and
it's
mostly
because
Colleen
is
trying
to
steal
some
stomach
texts.
To
add
to
that
vp9
we
discussed
now
will
change
the
will
update
the
the
milestone,
the
JPEG
X.
Still,
we
still
have
time
estate
and
the
TTL
is
already
done,
so
we
have
to
just
change
it
to
them.
A
E
F
In
a
cave,
I'm
sure
there's
a
lot
of
energy
around
you
transport
formats
and
potentially
media
over
those
new
transport
formats
and
whether
or
not
it
uses
the
same.
You
know
formats
as
abt
is
defined
or
not
is
still
an
open
question.
So
if
anybody
here
is
not
aware
of
that,
it's
just
a
strong
suggest.
You
follow,
like
you,
know
the
quick
work
and
the
rip
work
and
all
that
and.
A
C
Okay,
hi
I'm,
Colin
Perkins
I
want
to
talk
about
RTP
congestion
control
feedback,
so
the
draft
content
technical
content
has
not
changed
in
the
slightest.
In
this
revision
there
have
been
precisely
free
updates,
one
of
which
is
to
fix
my
core
Mallos
contact
details,
one
of
which
is
to
mention
the
REM
B
format,
along
with
my
minds,
blanking
timber,
and
that
sort
of
thing,
the
other,
is
to
add
a
paragraph
which
discusses
a
relation
with
the
Hulman
draft.
C
And
essentially
all
this
points
out
is
that
you,
the
home
address,
adds
a
night
EP
header
extensions,
give
a
packet
sequence
number
and
then
reports
that
as
a
sequence
number
for
the
aggregates
in
the
reports-
and
this
uses
the
sequence
numbers
which
are
already
in
the
packets
and
reports.
The
SSS
here
in
the
per
flow
sequence
numbers,
and
it
just
points
out
the
obvious
thing
that,
if
you're
adding
a
per
packet
sequence
number,
this
adds
you
overhead
per
packet
but
simplifies
the
reporting.
C
If
what
you
want
to
do
is
can
get
congestion
control,
the
aggregate
and
the
approach
in
this
mechanism
and
has
less
overhead
per
packet,
but
has
slightly
more
complex
reporting
and
gives
you
makes
it
easier
to
congestion
control
the
individual
flows
and
perhaps
also
integrates
better
with
the
rest
of
RTP,
which
also
reports
sequence
numbers
per
SSRC.
So
hopefully
this
is
a
sufficiently
neutral
description
of
the
trade-offs
and
will
not
cause
great
objections.
Hopefully
that
addresses
the
action
item.
C
A
A
A
Any
problem,
one
yeah,
basically
I-
guess
we
can
do
them
both
or
I've,
been
the
same
considerations
apply.
We
want
to
avoid.
You
know
the
holiday
season,
I
think
but
oh
yeah
well,
I
must
I'm
not
unless
we
think
we
get
about
a
fortnight
between
holiday
season,
right,
yeah,
so
so
I
guess,
but
so
probably
overlapping
this
with
very
marketing.
It's
fine
I,
don't
think
it's.
Neither
of
them
are
terribly
long,
so
nothing
either
of
them
will
require
yeah.
C
F
On
the
prior
note
about
the
new
trance
performance,
Thank
You
c'n
for
writing
all
this
up
concisely,
because
I
think
this
will
be
very
informative
for
people
trying
to
design
those
new
formats
and
thinking
about
how
feedback
and
reliability
and
recovery
work,
because
it
helps
to
really
distill
what
is
needed
for
a
media
application.
Some
people
that
are
reviewing
those
new
protocols
should
review
calls
draft
in
parallel
and
understand.
What's
missing
in
the
current
transport
drafts,
yeah.
C
F
Iii
would
greatly
doubt
that
those
other
groups
are
paying
attention
to
something
like
this,
because
media
is
kind
of
worked
on
their
side
right
now,
so
it
wouldn't
hurt
to
ask
them
to
consider
it.
But
I.
Think
specifically,
if
you
ask
the
quick
folks
that
are
working
on
the
recovery
draft
to
take
a
look
at
this,
because
recovery
mechanisms
and
reliability,
mechanisms
and
acknowledgments-
and
things
like
that
are-
are
critical
for
media
and
they
should
understand
why
and
how
they're
critical
for
media,
which
this
draft
lays
out.
But.
C
B
K
A
A
F
C
H
K
A
F
I
mostly
agree
with
dread,
but
that
you
know
there's
a
few
things
that
are
like:
we
don't
want
conflicts
between
the
areas
like
I.
Think
that
you
know
the
quick
stuff
says
you
know
I
w10
everywhere.
Basically,
I
forget
:,
whether
or
not
your
draft
says
w10
is
okay.
This.
F
A
C
L
L
L
So
our
approach
to
really
see
pillow
format
designs
to
really
start
from
the
service
of
NIH.
Most
Peace
Corps
is
widely
deployed
in
and
generally
works.
Even
though
there
are
some
still
open
issues
and
some
stuff
we
should
not
exercise
at
all,
but
that's
our
starting
point
most
apples
to
the
changes
in
our
header.
L
Our
current
first
strapless
doesn't
reflect
the
newest
to
change,
but
we
were
a
petition,
wants
the
general
meeting
in
GV
at
2020,
so
our
design
press
was
really
trying
to
remove
as
much
accomplished
as
we
can
base.
Our
studies
allows
for
things
in
cement
now
is
not
applicable
at
all.
Okay,
so
that's
a
way
of
trying
to
go
for
with
a
clean
design.
L
So
our
first
step
is
all
to
trying
to
do
B
this
transport
player.
First
right
and
all
the
silicon
is
somebody
gonna
come
last:
it's
not
because
we
forgot
it
it
better.
We'll
do
that
later,
so
we
targeted
this
thing
right
after
a
baby
see
our
first
version,
which
is
next
to
an
Atari
after
July
you'll
Ambridge.
So
that's
pretty
much
the
overview
really
see
if
any
questions
just
move
forward.
B
J
The
the
ITU
is
system
layer
standards
are
pretty
much
dead,
so
I
don't
expect
that
the
dependency
we
had
in
the
past
to
323
and
things
like
that
well
be
an
issue
here,
we're
not
on
a
time
pressure,
not
an
artificial
time
pressure.
There's
discussion
over
in
3gpp
essay
4,
that's
actually
in
work
item
that
they
want
to
perhaps
reference
that
and
they
would
need
it.
But
you
know-
and
that's
not
so
that's
not
so
they
for
the
codec
itself
is
just
for
their
the
reference.
They
called
it.
J
L
So
this
is,
to
the
overview
were
very
over
you
with
the
opening
issues
that
we
have
identified
so
far.
I
have
a
a
Celina
for
each
of
them,
so
I'm
just
gonna
go
through
what
are
we
are
concerning
the
open
issues.
The
first
was
really
concerned.
Well,
concern
is
something
we
need
to
discuss.
The
two,
the
transport
mode,
Mrs,
T
and
M
are
empty.
L
Then.
The
second
is
really
discussion
was
we
need
for
the
pocket
package
anymore,
ma
BC,
which
is
defined
in
the
semi
semi,
negate
that
they're
otherwise
really
c4?
We
need
this
answer.
Building
support
coming
from
this
dolly
basis,
signaling
for
the
incident
supporting
from
7/7
and
I
ate
the
last
one.
We
hope
you
can
leave
the
two
chairs
if
we
can
Japanese
working
group
draft
and
move
forward
to
sue
so
the
first
one
is
a
the
support
with
a
need
for
the
air.
L
Nor
stom
are
Machias
or
srst
is
simply,
everybody
knows
it
in
is
why
deployed
so
ii
would
say:
Mrs
T,
but
those
of
you
who
don't
know
what
is
is
very
carry
MOT.
Portuguese
stream.
Is
you
know,
transport,
the
distance
no7
I?
Don't
have
any
signaling
SST
B,
similarly
at
all
so
there's
a
long
history
to
be
back,
so
that's
even
longer
than
I
a
participant
a
standard,
so
I
will
leave
that
out,
but
that
every
settler
to
make
the
STP
signaling
for
Mrs
T
are
useful
in
our
eta
contacts
right.
L
G
L
L
So
this
miss
Mona
said:
Mrs
T
is
never
being
implemented
and
whatever
so
I
saw
the
questions
really
I
guess
we
don't
have
to
ask
this
question
anymore
I,
so
we
I
should
be
removed.
Just
now,.
A
F
Mostly
I
think
we
can
safely
omit
and
our
entity
I,
don't
believe
anyone
is
interested
and
in
that
I
grapple
and
forget
about
what
the
complications
are
with
Mrs
T
I.
Remember.
The
motivations
were
clear
that
you
could
you
could
more
seriously
detect
losses
and
things
like
that,
because
you
have
separation
of
the
layers
and
therefore
we
have
sequence
numbering
independent
for
all
of
those
layers,
but
I
suppose
that
with
something
like
frame
marking,
maybe
maybe
the
the
ambiguities
that
you
get
with
with
srst
aren't
as
bad
as
they
used
to
be.
F
But
I.
Don't
look
well
completely.
Shutting
the
bell
on
and
on
for
simplicity
on
floor,
eliminating
as
many
modes
as
we
can
I.
Just
wonder
whether
there's
anybody
has
actually
thought
about.
Are
people
actually
already
doing
the
marsh
team?
Don't
even
realize
it.
Certainly
simulcast
is
going
across
different
RTP
streams
right
nobody's,
bundling.
G
So
I
mean
Bernard
Abell
mic,
so
we
implemented
Mrs
T
for
h.264
and
one
of
the
arguments
is
you
don't
have
to
rewrite
this
sequence
numbers
so
there's
a
little
bit
of
an
efficiency
there,
but
a
lot
of
complexity
that
you
get
out
of
doing
it.
So
I
am
not
aware
of
anybody
doing
that
with
a
TV
see
it
is
done
with
h.264.
But
still
my
please
look.
J
J
J
We
had
references,
informative
references
to
drafts
and
stuff
in
there
in
the
hope
that
someone
gets
an
act
together,
ones,
one
certain
drafts
from
that
were
kind
of
clocked
over
in
RTC
web
once
those
drafts
were
were
were
done,
that
someone
would
get
the
act
together
and
do
abyss
or
something
no
one
ever
had
the
energy,
no
one,
no
one
cared
yeah.
That's
that's
not
only
because
of
the
success
or
lack
thereof,
of
HEV
C
and
in
the
video
conferencing
business,
but
also
no
one
cared
right.
J
No
one
cared,
it
I
hate
to
have
stale
text
that,
but
we
read
what
question
one
says:
should
we
do
it
now?
Should
we
wait
for
a
companion
document
to
do
it
later?
If
anyone
cares
or
should
we
person
to
be
Papa,
do
it
the
same
thing
that
we
did
with
711
98
and
do
it
in
abyss
and
both
later
options
are
something
which
I'm
kind
of
him,
maybe
the
first
one
I
just
say:
no,
we
shouldn't
write.
L
A
L
Great
all
right
so,
environment
T
is
auto
question
right.
So
when
we,
when
with
that
and
then
the
last
one
so
Bella,
the
fourth
noise
also
come
all
right.
So
I
need
four
right
history,
so
I
don't
want
to
go
through
the
history
again.
So
the
one
of
the
uses
of
the
extension
is
really
for
chimp
earlier
structure
in
signal
they
actually
no
other
uses.
You
may
have
identified
correct
me
if
I'm
wrong,
but
we
would
like
to
hear
if
there's
any
other
usage
for.
G
So
this
one
is
a
little
bit
different
I
do
understand
the
argument
that
some
kind
of
RTP
extension
is
likely
to
be
the
way
that
that
SF
us
operate
generically.
But
you
know
the
thing
is
there
there
may
be,
there
are
situations
in
which
SF
use
will
parse
the
bitstream
and
kind
of
prohibiting
that
from
working.
You
know
basically
requiring
you
to
negotiate
an
arch
at
the
extension
to
function
doesn't
seem
quite
right
to
me,
and
there
are
also
other
things
in
here.
I
would
say
that
that
can
be
supported
as
part
of
this.
G
J
To
put
SF
as
a
information
ever
as
a
fuse
would
typically
need
to
do
something
smart
to
a
scalable,
bitstream
I
think
is
not
the
payload
header,
because
the
payload
header
is
encrypted.
The
right
spot
is
in
an
extension
to
the
frame
marking
Draft
right
where
that
thing
is
not
encrypted
right.
So
the
scope
of
that
thing
is
only
the
payload
format
and
so
far
so
so
I
mean
we
we
can.
We
can
volunteer,
please
ship,
this
time
frame,
marking
draft
yesterday
right
dependent.
J
So
let's
get
that
thing
out
of
the
door,
but
I
officially
volunteer
here
to
think
about
an
extension
mechanism
of
the
frame
marking
draft
for
generic
stuff,
where
we
can
possibly
make
a
be
too
happy,
if
not
a
be
one
where
we
can
possibly
make
make
put
what
this
type
of
stuff
in,
but
throwing
in
a
generic
extension
mechanism
of
the
payload
header.
When
we
don't
have
a
single
use
case
that
makes
sense
in
today's
world
where
media
is
supposed
to
be
encrypted,
I'm,
sorry
I,
don't.
B
J
Okay,
so
so,
in
other
words,
we
may
want
to
have
in
here
a
section
that
says
how
to
use
the
frame
marking
draft
for
the
absolutely
agree.
Thank
you
very
much
for
reminding
me
yeah.
That's
what
we're
going
to
do
and
we
volunteer
to
do
some
work
there
and
a
yeah
go.
Are
you
online
territory
for
you
to
grab
work
to
do
next?
One
yeah
yeah
go
it's
muted!
Okay,
yeah
go
well.
Probably
do
it
hehe
volunteer
to
do
some
work
there
he
wants
to
be
involved,
so
Yahoo
can
do
it.
F
Person,
anyone
fall
one
on
that.
Thank
you,
someone
for
bringing
up
the
distinction
between
having
things
and
payload
headers
versus
artsy
headers,
like
the
reiterate
that
in
the
new
transport
work,
all
of
his
stuff
is
going
in
an
encrypted
basically
tunnel,
and
so
we
would
lose
the
ability
to
have
authenticated.
F
M
N
So
I
I
think
more.
What's
that
some
as
write
in
that
direction,
but
I
would
think
that
the
model
more
like
what
perk
houses
is
actually
what
we're
discussing
is
that
yes,
you're
gonna
have
now
to
encrypt
the
layer
between
the
nodes
in
a
centralized
conferencing
structure,
and
things
like
that,
so
you're
really
talking
is,
is
maybe
more
like
in
Prague.
How
do
you
expose
the
information
when
you
have
the
payload
being
encrypted
and
and
then
I
mean
really
the
media
payloads
so
and
I?
N
Think
if
I
remember
correctly,
that
would
mean
that
you,
the
payload
part
of
RTP
skarbek,
is
enter
and
encrypted.
Why
not?
They
had
or
extension
students.
It's
it's
still
in
this
booking
good
contact.
I.
Think
that's
the
the
right
way
are
thinking
about
this,
but
I
I
wouldn't
see
a
need
for
necessarily
doing
this
in
the
in
the
base.
Spec
for
this
payload
format,
I.
F
Got
Magnus
the
highest
runner
case.
Is
his
media
nodes
being
able
to
access
this,
but
I?
Think
one
of
the
other
motivations
for
having
our
theatres
and
the
clear
was
not
necessarily
a
conference
bridge,
but
even
on
path,
elements
to
be
able
to
see
and
do
things
intelligently
with
this
and
of
course
you
lose
that
ability
with
work
doesn't
allow
beyond
bath
elements
to
see.
You
know
anything
about
the
header
extensions
if
it's
going
really
over
DT
less
than
that
over
RTP,
not
ridiculous
or
to
me,
if
it's
going,
miscibility
yeah.
A
J
I
have
two
comments.
One
comment
is
that
the
typical
lifetime
of
a
video
codec
is
about
ten
years,
we'll
see
whether
we
see
initial
deployment
of
all
this
modern
stuff
that
more
sites
within
the
lifetime
of
this
codec
yeah
I
mean
come
on
this
we're
not
the
fastest
organization
on
the
planet.
That's
number
one
number
two
and
that's
probably
more
important.
The
argument.
J
The
the
key
argument
for
all
this
encryption
stuff
is
or
hey
you
know.
One
of
the
key
arguments
for
the
frame
marking
draft
has
been
as
far
as
I
understand
it,
that
the
middle
box
that
sits
there
in
the
middle
of
the
network
and
is
operated
by
someone
else
who
is
who
is
not
supposed
to
see
the
traffic
itself
can
act
on
it.
J
You
still
need
to
have
some
stuff
on
one
hand
encrypted
so
that
the
metal
box
can't
see
say
the
video
signal,
the
compressed
media
signal,
but
on
the
other
hand,
it
has
to
see
some
controlled
information
like
this
right,
so
so
just
putting
them
in
in
all
in
the
same
kind
of
giant
bucket.
It's
either
encrypted
or
not
in
that
scenario
is
wrong.
Yeah.
J
B
L
G
O
H
J
You
think
it's
not
okay,
it's
formulated
the
answer
to
question.
Five
is
no,
because
we
I
don't
think
we
need
to
extend
the
payload
header
itself
ever.
However,
there
should
be
a
question
5.5,
which
is:
do
we
need
a
mapping
of
the
mechanisms
of
the
new
BBC
codec
to
the
frame
marking
draft?
And
the
answer
to
that
question
is
a
resounding
yes
and
then
there
is
a
third
question
whether
we
need
that
in
this
document
here
and
the
answer
to
that
question
is
also
resounding
yes,
good.
E
L
All
right,
so
that
will
be
the
last
ones
we're
thinking
about
removing
the
Donny
based
signaling
for
engineering
support
M.
So
that's
a
better
trend
for
the
better
codecs
such
as
ABC
ABC,
M,
also
also
HTML,
maybe
see
that
we
know
the
coup
de
selva
is
already
good
enough
to
know,
let's
argue
hole
for
the
success
for
error
concealment.
L
A
I
B
B
A
B
A
A
A
P
P
P
P
It's
used,
it's
specified
in
next-generation
emergency
services,
we
do
regular
sip,
VoIP
services
and
in
relay
services.
So
next
slide
please,
but
the
multi-party
real
context
is
not
specified.
It's
mentioned
in
many
specifications
that
we
need
to
do
multi-party,
but
it
has
never
settled
in
any
standards
document.
It
has
been
said
that,
well,
you
do
you
use
regular,
ought
to
be
mechanized,
but
we
need
to
settle
it.
So
everybody
knows
what
is
the
best
way
to
do
it
so
that,
therefore,
the
current
user
terminals
only
support
to
park
a
new
of
Vietnam
text
course.
P
So
what
we
need
to
go
from
two
particular
multi
pocket
is
that
we
need
to
identify
the
source
of
text.
We
need
receiver
to
separate
text
from
different
sources
and
we
don't
allow
we
shouldn't
delay.
The
text.
Presentation
should
still
have
the
flow,
and
since
there
are
many
implementations
already
after
doing
two
party,
we
need
to
have
a
fallback
method
for
mixing
when
you
have
received
that
is
not
capable
of
sorting
out
the
receives
three
in
two
separate
presentations.
P
One
way,
which
I
think
is
becoming
the
most
common,
is
a
view
like
this,
where
you
labels
in
front
of
text
chunks
from
different
parties,
you
see
probably
the
three
lowest
icons
here.
You
have
people
typing,
so
the
text
is
going
from
multiple
parties
in
the
one
you
have
completed,
something
that
can
be
called
the
message
you
let
it
flow
up
in
the
in
the
next
slide.
We
have
another
way
of
displaying
a
real
context,
that
is
to
have
a
column
for
each
user.
P
K
P
It
should
not
delay
display.
We
have
an
overall
that
display
shouldn't
happen,
not
more
than
500
milliseconds
after
sending,
then
we
also
have
a
slight
risk
for
text
loss
when
we
are
redundant
same
use
so,
and
we
have
a
standard
saying
how
we
shall
indicate
that
to
the
receiving
user,
but
the
mark
in
the
stream-
and
we
need
to
have
that
still
available
to
mark
then,
and
we
shall
collect
text
from
different
users
so
that
it's
readable
and
the
source
is
identified.
P
We
also
support
erasure,
and
that
complicates
the
display
for
the
multi-party
case.
Even
must
not
let
it
erase
other
parties
text,
for
example
the
win
it.
What
for
security,
and
we
should
not
have
something
that
is
still
complicated.
There
is
a
some
urgency
for
getting
this
done,
so
implementable
in
near
future
would
current
implementations,
and
we
need
also
to
describe
this
for
back
method
for
cases
when
we
only
have
a
two-party
capable
of
receiving
terminal
mix
like.
B
P
P
And
since
they're
already
implementations
out,
we
need
to
have
a
method
for
capability
negotiation
for
the
multi-party
and
there
are
a
couple
of
methods
to
select
from.
We
need
to
discuss
how
we
identify
the
source.
We
have
some
presentational
aspects,
we
have
the
discussion
of
robustness
and
the
indication
of
loss
and
performance
and
security
and
Ayana
considerations
next
slide.
Please.
P
P
Now
they
drove
to
touches
topics
of
money,
working
groups,
it's
for
some
marketing
aspects
for
a
week
or
its
notion
ago.
She
ation
for
her
music.
It
might
be
something
Pacific
or
I,
don't
know,
and
so
we
need
to
arrange
the
discussions
in
a
practical
way.
I
wanted
to
take
advice
on
that,
so
we
have
it
in
a
music,
even
if
it
touches
every
key
topics
or
shall
they
distribute
the
discussions.
Let's
comment
and.
P
P
Examples
from
the
draft
some
discussion
topic
how
we
coordinate
RTP,
it's
different
marketing
methods,
translator
or
mixer,
more
have
in
the
source
indication
in
the
stream
or
have
a
mesh
mesh
about
the
end
points
or
even
multiple
RTP
sessions,
and
we
also
have
the
discussion
on
the
conference
on
the
web
with
agents
or
we
provide
some
sense
of
multi-party
to
them
next
night.
Please.
P
Another
discussion
topic
in
the
draft
is
the
capability
negotiation
where
I
have
identified
three
possibilities.
One
is
to
somehow
make
it
implicit
by
saying
that
if
you
have
a
multi
part
of
our
device
acquiring
text
in
the
stream,
it
must
be,
of
course,
speak
markup,
a
party
where
also
for
the
text-
and
you
can
start
sending
multi-party
contents
to
it.
Another
is
to
a
new
media
tag
for
declaring
multi-party
the
paper
lately
and
third,
one
is
to
have
STP
media
attribute.
P
Now
there
are
pros
and
cons
for
this
three
and
mark
the
other
ways
to
indicate
capability
next
slide.
Please
so
I
think
guy
Jeff
is
the
right
place
to
do
this,
because
we
have
needs
from
different
pockets.
We
have
Nina
and
we
have
three
EPP
dollars
and
they
often
refer
to
our
ITF,
currently
I.
Think
it's
a
best.
Current
practice
document
should
be
produced.
P
And
it
is
a
bit
urgent
to
get
this
ready.
The
urgency
has
been
announced
from
Nina
and
the
next
generation
emergency
services,
so
at
least
only
2021
that
I
want
to
see
implementations
running
Ollie's
and
max
like
this,
which
is
lost.
I,
look
forward
to
rapid
progress
on
this
and
hope
that
we
can
get
discussions
going
Thanks.
B
Okay,
so
I
think
we'll
have
to
go
and
discuss
with
the
a
dealer
to
progress
it
because
I'm
not
even
sure
if
it's
him
and
music,
because
it's
like
it,
has
a
spectrum
shape
like
we
said
from
c4n,
music
and
some
abt.
So
typically
we
do
it
through
dispatch,
but
then
I'll
consult
with
the
with
the
ad
about
what
to
do
and
we'll
come
back
and
saying
how
to
how
we
can
progress
it
and
whether
it's
well
and
then
we'll
have
to
decide
afterwards.
B
I
E
E
Taking
dispatch
is
a
reasonable
thing,
but
if
there
are
two
pretty
clear
working
groups
that
it
would
go
in
one
or
the
other,
and
you
should
work
that
out
first
and
if
the
answer
is
neither
of
them
thinks
it's.
The
answer
is
clear,
then
we'll
take
it
to
dispatch
or
I'll
chat
with
Adam
about
it
and
we'll
see
where
we
go
and
Magnus
obviously
has
something
to
say
so.
N
So
I
actually
have
a
question
which
I
think
defines
part
of
the
scope
to
goon
r8.
Are
you
feeling
that
the
multi-stream
multi-part
solution
is
clear
for
this,
that
the
current
payload
format
UNS
use
it
for
multi-party
with
when
you
actually
have
multiple
offices
are
sees
multiple
RTP
streams?
Is
that
clear
enough
in
the
specifications.
N
P
N
Mean
if
I
take
the
RTP
mobile,
the
obvious
multi-party
model
for
real-time
text
is
that
each
each
you
shoulda
tried
something
has
its
own
as
a
source
II.
The
stream
is
forward
the
whole
way
to
all
receivers
and-
and
it
becomes
solely
question
of
how
do
you
display
these
users,
it's
an
implementation.
It's
it
static,
fully
click
the
right
assumption
here
for
what
you
would
say,
be
the
right
solution
for
updated
users,
etc,
and
that
that
would
be
the
right
and
then
it's
the
question
of
how
do
we
deal
with
legacy
implementations?
P
N
Lot
of
people
spective,
say
I,
know,
I
know
the
result
is
this:
we
get
into
getting
the
whole
bundle
and
multiple
and
all
this
you
get
a
lot
of
em
lines
and
and
which,
which
is
interesting
in
these
cases,
where
you
have
I
would
say,
large
groups
of
conferencing
etc,
and
that
is
using
and
I
think
this
is
downside
of
the
result.
All
of
so,
therefore,
it's
an
M
in
some
sense,
let
me
use
the
question
of
the
complications
of
using
RTP
as
intended,
for
this
use.
N
E
I
N
So
I
mean
it's.
This
is
tough
I
think,
because
some
of
these
solutions
to
these
legacy
problem
is
is
in
the
space
that
you
actually
or
define
the
internal
of
the
text.
Formats
and
and
the
rest
is
very
much
tingling
I
think
for
dealing
with
et
cetera.
So
it's
it's
I
think
there's
a
large
part
which
is
probably
more
Indian
music
country,
and
then
you
have
this
specific
media
how
to
deal
in
media
and
that's
section
and
I.
It
might
be
that
it's
small
enough
and
it's
not
really
changing
the
RTP
stuff
anyway.
N
B
P
P
K
K
You
might
want
to
take
a
step
back
or
you
might
want
to
shut
down
the
scope
of
you
so
that
it's
okay
to
first,
you
a
point
solution
that
you
can
get
to
quickly
because
you
don't
consider
consider
either
other
requirements
or
other
technologies.
But
this
is
nifty
party
chat.
It
will
have
all
the
problems
of.
B
Trying
to
say
on
you
had
multiple
artists
solution,
architecture
in
the
document,
and
you
have
to
figure
out.
We
have
to
figure
out
which
one
will
be
simple
to
implement
and
to
start
with
that,
because
I
assume,
maybe
when
you
having
a
conference
and
a
conference
of
an
MCU,
the
does
that
may
make
it
simple,
because
signaling
would
be
simple
because,
like
a
point-to-point
with
handling
all
the
mechanism,
we
have
today
to
do
multiple
conferencing,
like
the
focus
the
one
to
a
defining
sync.
F
What
I
wanted
to
ask
is
a
qualifying
question
is
and
all
of
those
deployments
are
they
real
or
just
purely
nothing
but
text?
Well,
are
you
sometimes
trying
to
mix
the
text
with
other
media?
That
would
be
the
only
rationale,
but
I
could
see
why
you
would
go
to
any
kind
of
Artemia
solution
if
it's
purely
text
than
all
these
deployments
and
no
other
media
type
I
think
has
to
be
a
hack.
It's
an
accident
worse
than
an
accident.
It's
a
it's
a
giant
mistake.
Oh
yeah
I
can.
E
G
Bernard
about
myself
a
clarifying
question:
I
believe
this
is
part
of
the
next
gen
9
1
1
architecture.
Is
it
not
because,
as
I
yeah
recall
right,
you
can
have
real-time
text
and
audio
and
video
and
it
can
be
conferenced.
That's
how
the
recording
works
kind
of
go
for
a
conversing,
so
it
tip
it
can
include
both
audio
video.
A
K
P
A
Ims,
yes,
I
mean
I,
can
just
I
mean
which
I
guess
that
might
quite
there
is
that,
what's
what
implement
solution?
Spaces
are
likely
to
be
easy,
for
that
community
depends
yeah
that
effects
what
what
we
choose
to
do.
Yes,
yeah
so
I
think
it
does
sort
of
seem
like
people
are
feeling.
This
is
more
in
my
scope
with
M
music
than
here.