►
From YouTube: IETF-MOQ-20221021-1500
Description
MOQ meeting session at IETF
2022/10/21 1500
https://datatracker.ietf.org/meeting//proceedings/
A
Hi
folks,
thanks
everybody
for
being
here,
we
will
get
started
in
just
a
moment.
We
are
expecting
a
couple
more
people
at
the
very
least,
some
more
of
the
people
who
sent
us
inputs
like
Luke.
So
in
the
meantime,
I
will
start
the
administ
trivia
question
by
asking,
if
there's
anybody
who's
willing
to
take
notes.
A
We
really
do
need
a
note
taker
remember.
This
is
not
a
he
said.
She
said
style
note-taking,
it's
just
record
any
decisions
that
get
made.
So
in
this
case
we'll
probably
have
the
decision
on
which
style
of
GitHub
slack
yes
or
no
and
recording
which
personal,
which
technical
questions
to
tackle
and
the
incoming
personal
drafts
so
shouldn't
be
onerous
notes,
just
seven
or
eight
lines,
but
we
really
do
need
somebody
to
do
that.
A
A
Okay,
thank
you
James.
If
anybody
can
help
James
out
James,
will
you
be
using
the
note
taking
tool
here
or
not?
Yes,.
B
I'll
I'll
use
the
the
notes
of
the
thing
post,
thingy.
A
C
Is
calling
here
I
will
try
and
help
as
well
but
I'm
having
spotty
internet
connection,
so
I,
don't
I
can't
be
the
prime
person
thanks.
A
Okay,
so
the
next
thing
is
to
remind
you:
this
is
our
very
first
working
group
meeting
as
the
media
over
a
quick
working
group
as
opposed
to
a
buff,
and
so
it's
a
time
to
remind
you
of
the
note.
Well,
there
are
various
ietf
ITF
policies
which
it
calls
to
your
attention,
the
set
of
BCPS
being
at
the
bottom.
A
So
you
are
in
fact
expected
to
follow
the
anti-harassment
procedures,
the
code
of
conduct
and
to
be
aware
of
both
our
copyright
and
our
patent
policies
as
part
of
the
iitf.
If
you,
if
you
have
any
questions
about
those,
please
understand
that
once
you
start
participating,
we
will
expect
you
to
follow
them.
So
you
can
be
watching
if
you're
a
little
bit
confused
or
need
to
go
talk
to
a
lawyer,
but
please
don't
contribute
unless
you
agree
to
these.
A
If
you
do
agree
to
them,
please
contribute
as
much
as
you'd
like,
and
that
brings
us
back
to
the
second
bit
of
administrivia,
which
is:
is
there
anybody
who
needs
to
bash
the
agenda.
A
Okay,
hearing
No
Agenda
bashes
and
seeing
nobody
in
the
queue
we
will
go
on
to
our
Logistics
questions.
The
first
of
those
is
around
GitHub.
A
So,
as
people
remember
from
the
mailing
list,
discussion,
RFC
8874
discusses
a
couple
of
different
modes
in
which
a
working
group
can
operate,
so
some
of
them
are
issue
tracking
mode
and
another
is
issue,
discussion,
mode
and
I.
Guess
the
biggest
question
for
the
working
group
at
this
point
is:
do
we
want
to
use
it
at
all
and
presuming
that
we
do
which
of
the
modes?
Are
we
simply
storing
the
documents
there?
A
Are
we
using
the
issue
tracker
there
to
to
to
raise
issues
but
discussing
them
on
the
mailing
list,
and
or
do
we
want
to
have
some
of
the
issued
discussion
in
the
tracker?
The
difference
between
having
some
of
the
issue
discussion
in
the
tracker
and
having
it
on
the
mailing
list
is
primarily
how
early
the
mailing
list
gets
involved
in
the
decision.
Making.
The
mailing
list
is
always
the
place
where
the
current
version
of
a
document
is
approved.
A
D
The
the
one
thing
that
I
would
say
in
addition
to
what
Ted
was
saying
was
that
I'm
expecting
that
the
okay
repository
activity
for
any
of
the
repositories
that
are
involved
here
are
being
reflected
to
the
mailing
list
weekly.
So
it's
not
like
it's
not
like
the
people
in
GitHub
are
invisible,
you're,
just
not
getting
notification.
B
A
Okay,
I
see
a
comment
from
kiril
in
the
chat
saying
no
mailing
list.
Please,
that's
actually
not
one
of
the
options
as
Alan
points
out
the
mailing
list
is,
is
definitive
per
ITF
process.
A
A
Okay,
that's
a
little
bit
less
discussion
on
this
than
I
was
expecting,
but
maybe
everybody
just
agreed
with
everything
that
Alan
and
sorry
that
Spencer
and
Victor
said
so.
The
the
upshot
of
what
I
hear
is
there's
nobody
who
is
advocating
for
working
without
GitHub
and
that
both
of
the
people
who
spoke
positively
were
doing
it
in
the
mode
where
issue
tracking
is
done
primarily
in
GitHub
issues.
A
A
A
Okay,
note
takers:
please
put
that
down
as
the
discussion
to
be
referred
to
to
be
confirmed
on
the
list.
A
The
next
piece
of
Logistics
is
the
question
of
slack
yes
or
no
so
I'll
point
out
two
things
before
we
get
this.
This
discussed
one.
This
is
there.
There
is
no
ietf
document
that
documents
how
to
use
slack.
So
it's
it's
very
much.
A
An
ad
hoc
process
by
any
individual
working
group
I
think
the
working
group
that
uses
it
most
extensively
is
quick
with
a
quick,
Dev,
slack
and
I'll
point
out
there
that
the
quick
Dev
slack
has
one
serious
issue
for
some
members
of
that
working
group,
which
is
the
default
slack
client.
The
unpaid
version
of
it
only
keeps
a
certain
amount
of
History,
so
you
pretty
much
either
have
to
pay
for
it
or
be
aware
that
the
history
is
always
incomplete.
A
B
Just
we
once
again
follow
what's
a
quick
working
update,
which
is
slack
yes
for
interop
and
implementation
discussion,
but
generally
no
for
design
decisions,
and
it's
not
just
because
the
free
version
of
slack
is
not
archived,
but
even
if
it
is
Media
slack,
inherently
hard
to
search
things.
Inside
of
so.
If
we're
going
to
have
a
serious
design
discussion,
it
should
be
somewhere
that
is
trivially
searchable
and
linkable.
C
So
there's
many
ways
we
could
solve
this,
but
I
don't
think
we
should
have
I
mean
I'm
all
for
using
slack
or
a
tool.
I,
don't
really
like
slack,
but
some
tool
like
that,
like
that,
would
be
fine
I
think
we
should
use
a
tool
like
that.
However,
I
don't
think
we
should
use
any
tool
where
a
hundred
percent
of
it
does
not
end
up
in
a
long-term,
publicly
searchable
archive
that
ITF
can
use,
because
that's
critical
for
our
legal
process
I
strongly
object
to
things
other
than
that.
C
So
that
might
happen
just
due
to
you
know
being
mirrored
onto
an
email
list
or
something
like
that.
Like
the
same
way
we
do
with
the
PRS
or
with
the
the
get
the
GitHub
stuff,
but
I
think
one
way
or
another.
There
needs
to
be
a
long-term
public
Archive
of
this
stuff,
because
it
turns
out
to
be
really
important
later
when
there's
a
patent
argument
and
somebody
patents,
something
that
the
working
group
did
and
the
working
group
wants
to
show.
C
A
D
Thank
you
before
I
start
talking,
so
I
asked
this
I.
Think
I
was
the
one
that
asked
this
question
and
I
agree
with
what's
been
said
so
far,
the
the
one
thing
that
I
would
add
at
this
point
is
I'm
having
the
impression
that
we're
planning
to
do
with
more
with
zulup
in
the
future
than
we
are
with
slack
as
an
ITF,
so
we
might
just
want
to
keep
an
eye
on
whether
that's
a
better
answer
than
slack
as
a
the
free
version
of
slack
as
it's
used
today.
C
E
To
kind
of
qualify
that
question
saying:
can
we
use
slack
for
like
what
Victor
said
interrupt
and
implementation
like
clarification
discussions,
but
not
design
discussions
and
outcome
for
of
the
working
group
decides
what
to
do
next,
it's
very
easy
to
kind
of
get
into
a
mode
where,
as
part
of
the
discussion,
we
are
all
enthusiastic
and
we'll
start
this
design
discussions
there.
We
need
to
somehow
make
sure
that
it's
some
reflector
or
you
know
archived
so
that
you
can
be
searchable
at
some
point.
A
F
Okay,
yeah,
so
chair,
hop
off
for
a
second
I,
just
want
to
say
that
I.
Second,
a
lot
of
the
comments
about
having
for
interop
and
the
sort
of
like
technical,
real-time
discussion
for
quick
and
other
groups
has
been
super
useful,
so
I
think
having
a
channel
like
that
is
important.
I
also
want
to
just
bring
to
folks
attention
that
inside
the
quick
Dev
slack,
there
is
an
moq
Channel.
Now,
where
there
is
some
coordination
happening
between
the
authors
of
individual
drafts
there.
F
D
I
wish
I
was
just
going
to
ask
Alan
if
you
could
send
a
pointer
to
the
current,
wherever
you
think
the
actual
Place
starting
Monday
morning
would
be
for
for
that
to
the
mailing
list.
Thank
you.
F
D
I
know
that
you
and
Ted
will
do
the
right
right
thing.
I
think
what
I
was
mostly
saying
was
that
I
did
not
realize
that
such
a
thing
was
was
in
place.
D
So
if
you
know,
if
the
answer
is,
we
should
move
it
somewhere
else.
You
know
to
its
own
place
or
whatever
I
know
you
all
will
do
the
right
thing.
I
I,
just
don't
I,
just
don't
know
where
to
look
now.
Thank
you.
F
Yeah
I
understood,
I
think
that
the
draft
authors
sort
of
said
because
they
work
in
different
companies
and
we're
trying
to
iterate
rapidly
on
updates
to
their
drafts
prior
to
the
draft
deadline,
we're
like
we
need
a
Communication
channel
and
so
one
was
created
and
it's
you
know,
but
it's
it's
certainly
public,
but
do
not
maybe
easily
discoverable,
so
I'll
just
post
a
link,
maybe
even
the
minutes
of
this
meeting.
A
Well,
the
first
thing
we're
going
to
do
is
to
see
if
I
can
confirm
a
read
of
what
we
we
have
there.
It
seems,
like
everybody
agrees,
that
if
we
use
a
channel
like
this,
it
will
not
be
used
for
any
design
decisions.
A
It
would
be
used
during
interrupt
discussions
and
by
editors
or
or
implementers
for
quick
confirmations
of
their
understanding
of
what
the
spec
currently
says
so
I
think
we're
I
didn't
hear
anybody
argue
for
using
slack
for
anything
that
would
affect
the
design
of
the
of
the
protocol,
or
that
would
affect
the
the
long-term
specification
that
would
all
be
in
GitHub
or
on
the
mailing
list.
A
I
also
heard
pretty
strongly
concerns
about
the
The
Disappearing
nature
of
slack
in
particular,
I
I
believe
that
Cullen
Ali
and
myself
all
mentioned
that
so
I
believe
that
possibly
we're
not
looking
at
slack
itself.
So
what
I
propose
to
do
here
is
to
say
look
if
a
design
team
wants
to
use
a
channel,
then
it's
not
formally
part
of
the
working
group,
that's
cool!
A
What
we'll
do
is
go
off
and
see
if
we
can
find
either
a
mechanism
to
reflect
all
the
slack
messages
into
an
archival
medium
like
a
secondary
mailing
list,
or
something
like
that
or
a
digest
that
can
be
posted
to
the
mailing
list
or
we'll
investigate
using
zulip,
which
is
the
new
ietf
replacement
for
Chopper
and
which
does
have
this
long-term
archival
property
and
does
allow
us
to
create
multiple
channels
within
within
a
particular
Zula
topic
like
moq.
A
So
I
think
what
that
means
is
the
the
slack
answer.
Isn't
yes,
or
no
so
much
as
it
is
not
right
now
and
we'll
work
on
seeing
if
we
can
find
one
that
meets
the
archival
properties
relatively
quickly,
but
that
the
understanding
is
even
when
it's
in
place
it's
for
informal
Communications,
but
those
informal
Communications
would
be
retained.
C
Let's
see
I
guess
our
audio
seems
to
go
now
sort
of
okay,
but
I
I
think
you
should
probably
I
think
it'd
be
good
to
take
a
you
know,
get
some
input
up
from
the
room
on
zulup,
I
I
know
for
I
know:
I
definitely
would
not
want
to
use
at
all
I
just
find
it
very
difficult
to
use
I.
C
Think
there's
you
know,
Discord
or
slack
would
be
fine,
so
I
and
maybe
I'm
alone
in
that,
but
I
think
we
could
get
some
input
on
that
very
quickly
and
I
will
note
that
every
group
that
I
have
said
say
they
will
not
do
design
decisions
on
whatever
that
tool
is
promptly,
goes
off
and
does
tons
of
their
design
decisions
on
that
so
I
think
we
need
to
be
a
little
bit
more
I
mean
I
know
that
we
always
pay
lip
service
like
it's.
C
Just
like
saying
we're
going
to
confirm
it
on
the
list,
but
I
I
think
that
actually
we
need
to
be
sort
of
realistic
that
design
decisions
will
happen
on
this
stuff.
You
know
I'm
not
trying
to
change
people's
take
on
that
I'm.
Just
noting
that
it
happens
and
that
we
should
keep
that
in
mind,
When
selecting
a
tool.
A
Oh
okay
and
I
think
that
that
reinforces
the
need
for
the
archival
record
of
it.
Even
if
we
think
it's
going
to
be
informal
just
in
case
it
becomes
a
place
for
some
critical
piece
of
data
that
affects
a
design
decision.
F
Those
would
always
be
reflected
then
into
issues
in
GitHub,
with
discussion
happening
there
afterwards
and
I,
don't
think
by
any
means
that,
like
slack,
became
the
last
word
so
I
sort
of
expect
that
the
same
process
and
as
Spencer
would
say,
people
will
do
the
right
thing
here,
but
I'm
not
arguing
against
that.
You
know
having
a
long-term,
archival
and
searchable
history
isn't
is
also
important,
so
foreign.
A
Sorry,
does
anybody
want
to
talk
to
the
point
about
zulip
that
that
Cullen
raised.
A
D
I
I
think
I
do.
But
could
you
could
you
State
the
point,
the
killer
race
that
you're
asking
for
input
on.
A
Sure
Cullen
said
that
he
was
concerned
about
us
saying
that
we
would
look
I
either
how
to
take
things
out
of
slack
so
that
they
were
archived
or
we
would
look
in
zulip
since
it
seems
to
be
the
ITF
replacement
for
chapper,
and
his
comment
was
that
he
he
felt
it
would
be
wise
for
us
to
kind
of
pull
the
room
on
usage
of
zulip
because
he
felt
like
he
did
not
currently
himself
use
it
and
wasn't
particularly
keen
on
doing
so.
So
I
think
that's
a
a
question
here.
A
He
mentioned
Discord
as
well
as
another
option,
so
we
can
go
and
look
and
say
we
could
go
and
set
up
a
Discord
for
this.
It
has
the
same
sort
of
Disappearing
properties
for
some
people,
but
we
have
to
be
a
little
bit
aware
of
of
the
setup
of
too
many
tools
if,
if
people
are
using
slack
for
quick
and
Discord
for
Mock
and
zulip
for
HTTP
Biz
we're
we're
kind
of
all
over
the
map
there
and
that's
a
bit
hard
to
track.
D
Okay,
yeah,
okay,
so
the
prize
working
groups
can
do
the
right
thing
also.
So
whatever
the
right
thing
to
do
is
the
the
one
the
one
suggestion
I
would
make
about,
that
is
I,
don't
have
a
huge
amount
of
experience
with
Sula
I've,
mostly
used
it
during
medeco
meetings,
which
you
know
what
we
haven't
had
many
of
and
a
couple
of
months.
D
So
also
you
know
that
that
might
be
an
invitation
for
people
to
go.
Take
a
look
at
Google
up
and
see
what
the
what
they
think
about
that.
Also
a
reasonable
thing
to
do
would
be
to
ask
the
tools
team
if
our
understanding
of
the
Direction
with
zulup
matches
their
understanding
all
right.
If
I
remember
correctly,
we
had
a
test
with
zulub
and
some
other
similar
mechanism
that
they
they
pick
zulup
over
the
other
mechanism.
D
So
you
know
it's
like
I
think
they
have
a
direction,
but
just
for
us
to
make
sure
that
we
know
what
it
is
and
that
they
haven't
changed
it.
Thank
you.
A
Okay,
I
I
believe
that
the
Tulip
is
the
direction
we
can
confirm
that
it
hasn't
been
changed
and
I'll
take
the
action
item
to
do
that.
In
the
meantime,
may
I
ask
somebody
to
take
the
action
item
to
see
if
they
can
find
a
mechanism
that
would
allow
us
to
create
a
digest
or
digested
archive
of
a
slack
Channel,
and
if,
if
we
can
identify
one
of
those,
then
the
major
objection
to
using
slack,
which
is
that
it's
a
disappearing
medium
for
the
free
client
would
go
away.
A
A
A
Okay,
not
seeing
any
volunteers
for
this,
the
chairs
will
take
the
action
item
to
reconfirm
the
direction
on
zulub
and
ask
around
about
slack,
but
we'll
we'll
take
into
account
the
fact
that
the
this
seemed
to
be
significantly
lower
priority
than
the
GitHub
discussion
and
go
from
there.
A
A
Okay,
hearing
nothing,
we
will
move
on
to
the
use
case
document
I,
believe
James.
You
said
you
would
want
us
to
to
share
from
our
side
from
the
pre-loaded
slides.
Or
did
you
want
to
do
that.
A
D
A
I
haven't
actually
seen
the
request
yet
Alan
did
you
see.
F
D
D
I'm,
just
waiting
for
I'm
just
waiting
for
that
to
refresh
someplace
yeah
there.
Okay.
A
A
I
think
I
might
actually
have
to
advance
this,
then
because
it
looks
like
it
did
it
from
mine.
But
that's
okay,
just
tell
me
when
you
want
it
advance.
D
Okay,
thank
you,
so
I
think
James
and
I
are
both
talking,
so
we're
here
talking
about
a
individual
draft
that
we
had
worked
on
pre-charter
and
just
basically
what
what
to
do.
Next
with
that,
as
you
all
put
on
the
agenda,
that
sounds
like
a
next
slide:
Maybe
foreign.
D
So
the
working
group
is
chartered
to
adopt
a
use
cases,
a
requirements
document.
The
draft
we've
been
working
on
is
a
potential
candidate,
James
and
I.
Think
he
was
draft,
would
be
a
useful
starting
place
for
Bach,
but
recognizing
that
the
author's
pre-both
opinions
or
what's
in
the
draft
now
not
what's
in
the
approved
Charter.
D
D
Hearing
no
questions
could
we
go
to
slide
three.
D
So
we
think
so
we
had
started
on
use
cases.
We
thought
that
use
cases
would
drive
requirements
which
would
require
drive
work
on
protocols.
The
charter
includes,
but
is
not
limited
to
these
use
cases
quoting
live
streaming.
Gaming
and
media
conferencing
section
for
the
individual
draft
describes
these
three
use
cases,
and
so
we
had
three
questions
that
we
were
hoping
to
get
guidance
from.
D
The
working
group
The
chartered
working
group
about
whether
there
were
other
use
cases
that
are
in
scope
for
mock
that
have
unique
requirements
that
would
not
be
covered
under
live
streaming.
Gaming
and
media
conferencing.
D
I
I
think
it's
Ted.
Does
it
seem
reasonable
for
us
to
stop
after
every
question
and
let
people
talk
yeah,
you
can
ask
okay,
so
are
there?
Are
there
other
use
cases
that
are
in
scope
for
mock
that
have
unique
requirements
that
would
not
be
covered
by
the
use
cases
that
are
listed
in
the
charter.
D
The
oh
yeah,
absolutely
yeah,
absolutely
and
I
mean
the
other
thing
is
I,
don't
know
how
much
time
people
spent
looking
at
well
two
things.
D
Let
me
say
a
lot
of
the
work
on
this
was
done
before
the
the
trash
was
done
before
the
first,
not
not
a
working
group
forming
buff
on
Mock,
and
so
anybody
who's
been
paying
attention
since
then
may
not
be
familiar
with
the
draft
and
I
don't
know
how
much
time
people
have
spent
looking
at
the
slides
before
the
meeting
so
I
can
def.
I
can
definitely
ask
these
questions
on
the
mailing
list.
D
One
other
question
that
James
and
I
have
wondered
about
was
whether
the
Hybrid
used
case
that's
described
in
section
4.2
of
the
draft,
and
so
this
is
to
be
clear.
This
is
a
use
case
that
has
aspects
of
both
live
streaming
and
highly
interactive
gaming
media
conferencing,
kind
of
requirements.
B
And
and
the
example
that
that
has
been
cited
on
quite
a
few
occasions,
is
the
the
Staff
All
Hands
meeting,
where
there
may
be
hundreds
or
even
thousands
of
viewers,
but
only
a
small
number
of
active
people
being
interactive,
both
producing
sending
media
as
well
as
receiving.
And
it's.
Whereas
you
know
it's
the
vast
majority
of
purely
just
receiving
and
consuming.
D
E
Thank
you,
Spencer
for
bringing
this
I
think
some
of
the
discussions
with
the
unified
work
that
we're
also
trying
to
do
has
this
aspect
of
how
do
we?
Let
viewers
like
Luke,
would
also
talk
talking
about
it
more,
but
it's
more
about
how
do
we,
let
users
say
the
con
or
viewers
say
the
latency
they
would
expect
and
and
having
the
the
delivery
protocol
meet.
E
D
I
can
I
can
make
a
my
opinion
statement,
which
is
that
ITF
work
often
is
not
super
focused
on
use
cases
as
a
starting
point,
so
we
might
be
adding
some
material
in
the
use
case
portion
of
the
draft,
but
I
think
the
intention
right
now
is
to
have
the
same
document
with
use
cases
and
requirements
so
that
it
may
be
that
it
may
be
that
what
you're
talking
about
as
we
drill
down
our
requirements
would
would
be
more
naturally
part
of
that
section.
D
The
other
question-
and
this
may
be
getting
into
this-
may
be
getting
into.
D
Discussion
that
would
happen
after
you
know
we're
moving
forward
with
this
draft
of
the.
If
it's
okay,
if
the
working
group
moves
forward
with
it,
is
making
sure
that
we
have
enough
descriptive
material
in
the
use
case
section
to
explain
why
we
have
certain
requirements-
and
you
know,
as
as
suhas
was
just
was
saying,
there's
certain
require.
You
know
certain
use
cases
with.
That
would
say
this
is
you
know,
because
we
support
this
aspect
of
this
kind
of
use
case.
This
is
why
this
is
a
requirement.
D
So
that
may
be.
That
may
be
the
answer
to
our
third
question.
There
so
I
think
any
any
more
comments
on
this
slide.
D
And
if
there
are
not,
could
we
go
to
the
next
slide?
I
think
we've
I
think
we've
had
the
discussion,
so
there
is
one
more
slide.
What
James
and
I
are
planning
to
do
are.
D
Actually
we
look
at
this
far
is
to
update
the
abstract
and
introduction
to
reflect
the
approved
Charter,
there's
quite
a
bit
of
History
observations
and
our
opinions
throughout
the
draft
that
can
be
removed.
D
The
we
that
updating
the
use
cases
to
reflect
working
group
discussion
on
slide
three
questions
would
be.
You
know
the
other
thing
that
we
would
want
to
do
before
we
before.
D
We
asked
anybody
for
an
option
of
this
draft
by
the
working
group
and
in
proposing
an
initial
structure
of
the
requirements
section
based
on
implicit
requirements
that
are
in
the
approved
Charter
and
at
this
point,
if,
if
this
rolls
the
way
I
was
thinking
it
might,
the
working
group
would
be
controlling
both
the
use
cases
and
requirements
for
for
mock.
At
that
point,
any
comment,
comments
or
questions
on
this
Cullen.
C
C
If
there's
Parts,
you
know
delete
as
much
like
there's
a
lot
of
history
in
there,
let's,
let's
like
yeah
intense
it,
make
it
easy
for
everyone
to
read,
but
also,
if
there's
stuff,
that's
just
not
agreed
upon
on
the
working
group,
but
there's
a
range
of
opinions.
We
can
express
that
and
then
adopt
it
as
a
working
group
document
with
the
issue
not
resolved
but
an
issue.
You
know
we
have
a
range
of
options
we
could
go
here.
There
isn't
yet
working
group
agreement.
C
D
Guys,
yeah
yeah
yeah,
so
so
I
I
think
that
I
think
that's
I
think
that
there
are
things
in
the
draft
that
are
not
in
scope,
but
I've
not
done
a
quick
I've,
not
done
the
past
in
the
last
20
minutes
for
what
what's
in
there.
That
is
not
agreed
upon
yeah
that
obviously
agreed
upon,
but
that
is-
and
you
know
that
the
answers
would
be
in
scope
so
that
that
that
helps
me.
You
know
that's
a
very
helpful
comment
to
me:
Colin.
Thank
you.
D
E
Yeah
I
I
think
this
is
a
useful
document
to
have
and
it'll
help
us
drive
the
protocol,
design
and
architecture
as
we
go
forward
and
and
and
I
agree
once
once.
The
authors
are
ready
with
the
next
revision.
We
should
consider
it
for
adoption.
Thank
you
for
the
work.
B
I
think
there's
one
point
in
clarification
that
is
probably
worth
mentioning,
which
is
that,
upon
adopting
this,
the
our
draft
after
we
address
the
various
points
that
we've
listed
out
doesn't
necess,
doesn't
necessarily
mean
the
working
group
initially
agrees
with
everything
that's
in
there.
B
However,
once
it's
brought
into
the
working
group
is
adopted,
then
everybody
who
participates
in
the
working
group
can
then
use
the
consensus
mechanisms
to
then
get
some
of
those
things
fixed,
particularly
if
they're
somewhat
controversial
so
I
I
would
warmly
like
to
to
keep
for
folk
to
keep
that
in
mind
that
they
can
be
definitely
disagreeable
points
in
here
and
that
by
bringing
them
bringing
the
document
in,
we
can
then
argue
about
it
and
then
get
a
consensus
over
over,
and
you
know
this.
B
This
POI
has
been
done
by
other
working
groups.
It
just
so
happens
that
my
own
other
active
document
was
done
in
this
is
being
done
in
this
way
and
it
seems
to
work.
D
Yeah
and
and
James
James
says
for
about
what
James
said
is
reminding
me
my
interest
at
least
my
understanding
of
what
it
means
to
adopt.
This
is
that
the
working
group
thinks
that
this
is
a
document
that
is
a
reasonable
basis
for
working
group
work,
so
any
other
any
other
comments.
I'm
gonna,
I'm
gonna
have
a
question
for
the
chairs
and
and
the
working
group,
of
course,.
A
D
Yeah
I
I
was
not
really
thinking
about
when
we
did
when
we
did
the
when
we
did
the
first
version
of
the
slides
for
this.
For
this
talk,
I
was
not
really
thinking
about
adoption
now
and
you
know,
as
the
slides
reflect
but
I'm
wondering
you
know.
If,
if
you
know
Cullen
said
you
know,
he'd
be
okay
with
adopting
it
now,
with
this
work
plan
as
part
of
that
and
would
other
people
would
other
people
be
okay
with
that,
so.
A
With
chair
hat
on
since
that
wasn't
one
of
the
decisions
we
announced
would
be
part
of
the
the
interim
I
I
would
prefer
not
to
do
that
at
this
interim?
If
you
guys
produce
the
document
with
the
revisions,
we
can
take
that
up
in
London
in
London
for
itf-115,
but
I'm
I'm
a
little
bit
uncomfortable
with
doing
it
at
this.
This
interim,
we
could
do
it
on
the
list
as
well.
Yeah.
A
A
D
C
C
It's
not
working
group
consensus
for
the
parts
that
aren't
in
the
dock
before
we
adopt
it
for
for
working
group
consensus,
but
I
felt
bad
for
you
presenting
and
getting
zero
feedback,
and
it's
like
do
people
love
this
I
think
you're
on
the
right
track.
Please
do
the
work!
Thank
you
like
that's.
What
I
was
really
trying
to
say.
D
And,
and-
and
thank
and
thank
you
for
thank
you
for
clarifying
as
well
so
like
I,
said
what
you
just
said
is
kind
of
what
I
was
thinking.
We
were
going
to
be
doing
between
now
and
London,
I
think
so.
The
the
ID
cutoff
is
like
Monday
right
yeah,
so
so
it
is
not
super
likely
that
we
would
that
we
would
produce
and
submit
a
draft
by
Monday.
D
That
does
that
the
the
one
thing
that
I
also
wanted
to
follow
up
with
Port
Cullen
was
saying
was
there
are
things
that
we
could
we
might
or
might
not
take
out
of
the
draft,
but
the
things
that
are
not
part
of
the
working
group
Charter,
those
are
the
things
that
definitely
need
to
go,
and
then
you
know
then,
once
once
everything
we're
talking
about
is
in
the
under
the
current
Charter
I.
D
Think
that
would
be
a
fine
time
to
to
for
us
to
talk
about
the
working
group
adopting
this.
So
do
you?
Do
you
want
me
to
send
the
request
for
a
slide
in
London
now.
A
Sure
we'll
we'll
consider
a
a
request
for
a
slot
in
London,
based
on
the
on
the
minutes
of
this
meeting
so
perfect.
D
You
are,
you
are
a
sport.
Thank
you,
I
think.
That's
I,
think
that's
everything
that
I
know
back
to
the
rest
of
the
meeting.
A
Okay,
the
next
was
the
question
of
which
technical
questions
to
tackle
in
London
and
Sue.
How
scared
a
presentation
on
that.
E
Okay,
I
can
hear.
Can
everyone
hear
me
yes
or
not?
Okay,
great,
so
the
the
main
idea
behind
this
slide
deck
is
not
going
deep
into.
How
do
we
solve
each
of
the
things
that's
been
listed
in
the
slide,
but
basically
looking
at
the
charter
and
Distilling
it
one
level
down
to
see
what
kind
of
questions
that
we
would
like
to
answer.
It
might
turn
into
requirements
that
would
help
like
Spencers
on
James
document,
or
it
might
turn
into
a
task
list
that
the
working
group
would
want
to
follow.
E
So
keeping
that
context
in
the
mind
I'll
jump
to
the
next
slide.
Please,
okay,
topics
that
will
will
be
talking
about
here,
which
is
basically
rocked
from
the
charter.
We
need
to
work
on
a
common
published
protocol
between
ingest
and
distribution,
and
we
need
a
way
to
name
or
address
things
whatever
we
call
it
as
to
identify
things
and
so
that
we
can
request
them
appropriately
and
also
we
need
a
way
to
kind
of
package.
This
media
we
have
different
formats
there
like.
E
E
Published
protocol
required
it
needs
to
be
common
between
the
contribution
and
distribution
side
of
the
picture,
and
we
need
a
way
to
support
to
specify
different
kinds
of
media
types
like
it's
an
ad
audio,
video
or
subtitles,
or
it's
an
ad
or
something
and
also
different
qualities.
So
those
media
types
correspond
to
and
also
different
codecs.
Those
are
some
of
the
things.
At
least,
we
need
to
be
able
to
kind
of
specify
as
part
of
the
protocol
and
on
the
things
that's
being
published,
and
these
things
need
to
be
added.
E
We
need
to
be
able
to
identify
them,
uniquely
so
that
on
the
consumer
side,
you
should
be
able
to
kind
of
retrieve
that,
and
also
when
you're
publishing
the
media
can
be
published
at
various
granularity.
For
example,
if
you
take
about
the
Via
to
talk
about
the
video,
a
video
can
be
published
as
individual
individual
frames
that
comes
out
of
the
encoded
encoder
or
it
could
be
group
of
frames
like
the
group
of
pictures
that
we
usually
to
see
in
the
video
encoding
or
something
different.
E
We
need
to
kind
of,
say,
group
decide
on
on.
What's
the
right
way
to
go
about
it,
and
some
of
these
things
will
also
be
addressed
when
Luke
presents
about
unified
draft,
but
here
I'm
putting
it
here
so
that
we
need
to
address
this
technical
detail
next
slide,
please,
and
once
once
the
media
has
been
published
under
the
transport
in
order
to
make
a
relay
the
first
class
citizens
in
the
architecture.
E
We
need
to
have
some
kind
of
metadata
that
goes
along
with
the
media
being
published
that
helps
the
relays
or
the
caches
of
the
proxies
to
make
the
right
caching
decisions
like
either
you
want
to
drop
or
forward
and
how
to
do
it
under
congestion,
especially
in
the
media
media.
Agnostic
way.
E
We
don't
want
the
realized
or
proxies
to
be
worried
about
every
codec
that
comes
out
in
the
future
or
every
code
that's
been
built
in
the
past,
so
a
media
agnostic
way
to
kind
of
provide
metadata
that
can
be
read
by
the
relays
having
some
information
about
the
identification
of
the
media
being
published
the
priority
some
dependencies,
if
it's
there
and
also
should
we
cache
it
or
not.
E
Helping
this
kind
of
relays
or
the
middle
boxes
make
decisions
without
having
to
look
into
the
actual
media
payload,
which
will
be
encrypted
and
also
which
the
protocol
should
enable
to
configure
to
support
various
use
cases
that
James
just
meant
mentioned
about
live
streaming
or
interactive,
media
or
media
conferencing.
We
need
to
be
able
to
kind
of
specify
that,
in
the
protocol
and
and
whatever
we
send
should
be
able
to
map
on
the
quick
stream
or
quick
streams
or
quick,
datagrams
or
datagrams
is
something
we
are
for.
E
We
are
evaluating,
but
we
for
for
the
use
case,
especially
the
media
conferencing,
our
interactive
use
cases,
but
something
we
need.
The
protocol
should
provide
a
way
to
map
and
also
the
other
things
is
that
how
do
we
send
the
data
or
web
transport
or
HTTP
3
or
Rock?
E
Quick,
just
some
of
the
things
that
we
need
to
consider
when
designing
this
protocol
and
application
data
can
be,
can
be
end-to-end
encrypted
in
the
sense
that
from
the
publisher
to
Consumer,
if
you,
if
there's
no
transcoding
or
anything
happening
in
between
how,
if
you
support
or
something
like
that,
how
even
the
even
the
the
how
the
key
keying
happens,
everything
is
outside
the
scope
of
this
protocol,
but
we
need
to
kind
of
make
sure
that
the
the
data
in
transit
is
not
or
cannot
be
read
by
the
middle
boxes.
E
Next
slide,
please
that
kind
of
covers.
You
know
what,
on
the
publication
side
of
things
to
happen,
Okay
one
of
the
important
things
also
that
goes
along
with
either
publication.
That's
also
common
to
the
consumption
is:
how
do
we
name
or
identify
the
things
that's
being
published?
E
Given
we
are
working
in
the
web
model
with
cdns
and
kind
of
relays,
and
also
the
origin
in
the
picture.
We
need
to
make
sure
that
water
being
published
is
in
the
scope
of
an
application
domain
or
the
origin,
and
that
would
that
basically
means
that
a
somehow
it's
tied
to
a
billing
relationship
that
happens
with
a
company,
that's
kind
of
actually
producing
the
media.
It
might
not
be
a
direct
goal
of
this
protocol,
but
this
architecture
works
that
way.
E
We
need
to
kind
of
keep
that
in
mind
and
the
naming.
What
naming
basically
does
is
identifies
these
Elemental
streams
or
the
variations
of
those
things.
So
that
way,
it's
it's
very
easy
for
the
consumer
to
kind
of
request,
the
right,
quality
or,
or
or
or
the
consumer,
to
kind
of,
or
or
consumers
kind
of
pick
between
different
qualities,
all
right
and
for
the
final
scenario,
scenarios
from
the
CDN
and
to
multiple
subscribers.
We
are
talking
about
like
millions
or
like
hundreds
of
thousands
of
subscribers.
E
We
need
to
be
able
to
kind
of
May.
The
name
should
enable
the
consumers
to
kind
of
ask
for
the
data
they
need
that
matches
a
given
name.
E
Next
slide,
please
on
packaging
media,
while
we're
transporting
there
are
multiple
options
that
we
want
to
package
media,
some
of
the
contenders
that
we
have,
or
some
of
the
options
that
we
can
the
can.
They
coexist
or
something
like
using
existing
container
format
like
cmaf
and
also
supporting
Rock
container
format.
E
If
for
the
cases
where,
if,
if
we
think,
if,
if
CMS
system
cmap
does
not
does
not
help
their
supporting
a
raw
container
format
and
as
a
group
we
need
to
decide,
do
we
need
any
other
content
formats
that
that
would
help
make
that
option
of
a
mock
protocols
widely
next
slide.
E
Next,
moving
on
the
consumption
side,
so
we
we
once
we
once
we
kind
of
design
on
how
the
names
should
be,
how
on
what
being
published
and
what
they
identify
the
different
qualities
and
the
things
the
consumers
should
be
able
to
kind
of
ask
those
names,
and
also
consumers
should
be
able
to
kind
of
ask
the
right
things
based
on
the
names
based
on
their
names
like
each
name
should
identify
what
kind
of
media
it's
being
is
being,
is
being
sent
and,
and
also
the
quality
and
the
the
properties
of
that
media,
so
that
the
consumers
should
be
allowed
to
kind
of
ask
those
kind
of
things
and
also
stop
if
it's
no
longer
interested
and
again
since
we're
working
in
this
web
operation
model.
E
So
the
consumers
are
always
authorized.
That's
the
origin!
It
goes
without
saying
that
next
slide.
Please.
E
And
on
also
on
on
the
requirements
of
what
a
relational
caches
should
do,
the
the
basic
functionality
is
to
kind
of
they
should
allow
caching
and
distribution
of
the
media
data
as
applicable,
and
also
they.
They
basically
store
data
based
on
the
trust.
That's
been
set
up
by
the
origin.
Again
this
course
without
saying
that,
and
also
like,
like
how
the
publisher
and
consumer
kind
of
work
on
quick
streams
or
quick,
datagrams
or
or
whatever
we
come
beside,
the
protocol
should
be.
E
The
release
should
also
be
doing
that
and
for
cases
where
the
latencies
need
to
be
really
kept
low.
Some
of
the
options
like
pipelining
might
be
considered
where,
where
the
the
less
time
is
spent
on
on
the
chain
of
release,
so
that
N20
latency
can
be
reduced.
E
But
these
are
some
of
the
things
we
need
to
explore
as
we
design
the
protocol
and
relay
should
be
able
to
kind
of
make
their
cash
or
drop
decisions
just
based
on
the
metadata
and
nothing
and
and
not
being
able
to
access
the
actual
payload.
The
mock,
payload
and
I
really
should
be
able
to
kind
of
push
the
data
based
on
the
consumers
asking
for
the
data
push
the
data
whenever
it's
available
to
fan
out
to
the
any
number
of
consumers.
E
Next
slide,
please
on
end-to-end
security,
the
the
publication
protocol
or
itself
would
not
be
kind
of
is
out
of
scope
of
the
of
the
protocol
to
say
how
the
keys
for
the
encryption
would
happen,
but
we
need
to
kind
of
consider
possible
options
that
would
happen
to
support
the
use
cases
that
we
have
listed
is
something
like
a
DRM
based
like
the
the
content,
can
be
DRM
encrypted
to
support
many
of
the
use
cases
on
the
live
stream
side
of
things
or,
if
you
are
going
with
like
media
conferencing
use
case,
something
like
an
MLs
or
some
combination
of
s
frame
might
be
one
thing
that
we
might
consider
or
if,
if
people
think
there's
some
something
different,
that
should
be
good
to
let
good
mechanism
or
or
the
useful
mechanism
that
could
be
used
to
derive
the
keys
for
encrypting
the
content.
E
Something
like
that.
We
we
need
to
kind
of
talk
about
that
as
well.
I,
think
that
should
be
all
this.
This
kind
of
a
laundry
list
of
things
that
we
might
might
want
to
do
and
I
would
like
to
kind
of
encourage
working
group
to
kind
of
think.
Through
these
questions,
either
we
create
a
list
somewhere
where
we
can
go,
discuss
further
and
kind
of
see.
How
can
we
build
our
solution?
That
kind
of
fits
the
requirements
that
we
have.
F
Thank
you
has
so
I,
just
so
a
few
minutes
to
discuss
here
and
I
think
the
what
might
be
interesting
is
to
try
to
surface
the
topics
that
we
think
we
could
put
on
an
agenda
in
London
to
have
like
a
deeper
discussion.
There's
obviously
a
lot
of
topics
that
we
have
to
cover
in
this
working
group
over
time,
but
think
about
what
are
the
first
ones
that
we
want
to
really
dig
into.
That
will
maybe
have
more
Downstream
implications.
D
So
one
of
the
so
one
of
the
things
I
was
going
to
suggest
that
we
focus
on
early
is
getting
a
clear
specification
of
what
we
think
a
protocol
stack
looks
like
this
was
discussion
that
was
happening
in
the
email
thread
that
Luke
started,
and
there
were,
you
know,
there's
at
least
three
different
variations
that
I
can
think
of
that
people
seem
to
have
had
in
their
heads
when
they
were.
You
know
when
we
were.
D
We
were
talking
about
this
I
think
that
that
will
make
things
a
lot
clearer
for
a
number
of
these.
A
number
of
these
questions,
I
I,
was
looking
at
your
thing
on
the
end-to-end
security,
which
maybe
that
may
be.
D
That
may
be
a
good
topic
for
London,
and
the
thing
I
was
going
to
say
was
that
the
mops
document
on
operational
considerations
for
streaming
for
streaming
providers
has
a
security
section
that
talks
about
you
know
I'm
encrypting,
at
the
application
layer,
I'm
encrypting
hubby
hop
I'm
encrypting
in
end-to-end
and
kind
of
what
the
considerations
are
there
for
each
of
those
that
may
be
a
that,
may
be
a
reasonable
template
to
to
use
I.
D
Think
you
did
say
something
that
was
starting
to
make
me
wonder
we
you
know
we
have
things
that
we
want
to
make
sure
that
we
don't
break
existing
business
models
unnecessarily.
D
So
for
us
to
be
thinking
about
that
as
we're
meeting
in
London
and
I
wanted
to
ask
if
we
had
a
GitHub
with
a
GitHub
Wiki,
is
that
something
that
we
could
set
up
quickly?
To
start
this
catalog.
D
Or
is
there
a
better
way
to
do
it
because
I
think
there's
I,
think
you're
I
think
you're
putting
a
lot
of
stuff
in
front
of
the
working
group
that
all
of
it
matters
and
for
us
to
figure
out
how
to
how
to
how
to
do
that
in
another
huge
amount
of
time.
E
But
but
thanks
Spencer
valid
points,
I
I,
I,
I
think
having
having
a
GitHub
Wiki
where,
but
because
most
of
the
things
I've
listed
down
kind
of
is
a
digital
down
version
of
what's
in
Charter,
some
might
be
my
personal
opinions,
which,
which
is
definitely
open
to
change
having
this
as
a
backlog,
or
something
would
definitely
help
doing
that
and
I
also
agree
with
Alan.
What
Helen
said
is
that,
yes,
we
have
Charter
says
we
have
to
do
15
things.
E
What
is
the
first
three
things
you
want
to
do
and
kind
of
making
that
separation
saying
this
is
the
first
three
things
we
want
to
work
on
and
then
rest
would
come
eventually
follow
after
that
would
would
be
good
model
and,
at
the
same
time,
even
though
we
are
trying
to
design
a
protocol
that
works
both
on
the
production
and
also
the
consumption
side,
keeping
these
things
in
mind
would
help
us
not
go.
E
D
F
So
thanks,
Spencer
and
I
I
know
I,
hear
the
request
to
make
sure
we
don't
lose
track
of
this
list
of
things
that
we
know
we
need
to
tackle
so
I
think
Ted
and
I
can
make
sure
that
we
find
a
home
to
to
track
this
backlog.
F
Does
anybody
else
want
to
weigh
in
on
what
they
think
the
one
or
two
or
three
most
important
or
the
highest
priority
items
to
tackle?
First
are
besides
Spencer.
D
I
was
I
was
I
was
for
the
purposes
of
this
conversation,
I
was
trying
to
get
out,
I
I
can
I
can
I
can
I
can
always
say
something
else
and
if
nobody's
lined
up
behind
me
anyway.
D
Whatever
the
right
document
is,
and
it
may
be,
the
protocol
document
you
can,
you
can
replace,
you
can
create
the
repo
for
it
now
and
then
we
can
be
putting
these
in
as
issues
and
using
labels
on
you
know.
This
is
for
ITF
115.
This
is,
for
you
know
it.
You
know
this
is
for
later
kinds
of
things
to
to
manage
the.
So
we
don't
have
so
we
don't
start
out
with
500
issues,
but
just
just
to
just
to
give
us
a
place
to
get
started
and
get
organized.
E
I
know
if
I,
if
I
may
suggest
like
since,
since
one
of
the
first
deliverables
we
have
on
along
with
use
case
document,
is
the
the
common
media
publication
protocol.
If,
if
you
take
that
as
the
the
category
that
we
want
to
kind
of
solve,
first
and
and
and
some
of
these
things
will
move
out
or
move
out
automatically
from
from
from.
E
If
you
take
that
as
a
goal,
maybe
when
you
create
a
Wiki
or
somewhere,
we
basically
go
by
by
that
categorization
saying
that
we
start
with
common
media
protocol
media
publication
protocol,
and
for
that
these
are
some
things
we
have.
We
we
might
want
to
see
if
you
can
address
this,
these
technical
questions
for
those
things,
maybe
that
might
be
the
first
start.
F
Okay,
I
saw
a
comment
from
Luke
in
the
in
the
chat
about
prioritization,
but
also
referencing,
his
upcoming
slides,
so
I,
don't
know
if
Luke
you
want
to
jump
in
and
Echo
that
the
room
or,
if
other
folks
have
different
opinions
like
I,
think
what
sue
us
just
said
about
well
I.
Think
writing.
The
common
publication
protocol
is
like
the
higher
like
one
of
the
highest
priorities,
so
we
should
take
that
and
then
work
on
subtopics
from
there.
First.
H
H
Yeah,
the
most
important
part
of
a
media
transport,
or
just
a
network
transport
in
general,
is
making
sure
the
data
gets
there
correctly.
So
I
mean
it.
It's
really
tempting
to
start
like
nitpicking
like
how
do
we
support
DRM,
but
at
the
same
time
it's
like
how
do
we
make
sure
that
media
gets
from
point
A
to
point
B
in
reliable
fashion
such
as
it
can
be
decoded
and
there's
a
few.
H
The
few
different
drafts
out
there
disagree
a
little
bit
on
that
and
that's
why
I
wanted
to
kind
of
do
a
presentation
of
the
the
common
ground.
F
Okay,
thanks
does:
is
there
anybody
else
who
wants
to
oh
mo.
B
I
just
wanted
to
agree
with
Luke
that
packaging
of
the
media,
I
think,
is
the
most
important
thing
to
tackle.
First,
because
I
think
that
quickly
shows
the
differences
that
people
are
assuming
and
it's
it's
critical
for
the
transport
of
the
media
to
understand
the
the
protocol
boundaries
of
fragments.
B
F
Okay,
thanks
I
I,
I,
think
I'm
getting
a
sense
of
of
where
people
want
to
go.
Does
anybody
else
want
to
weigh
in
or
should
we
should
move
on
to?
The
next
will.
B
E
I
I
think
I
I,
second,
with
Will,
some,
the
thinking
there
and
and
I'm
going
with
like
what
Mo
and
Luke
was
saying
that.
How
do
we
distribute
this?
The
the
publish
the
media
from
point
A
to
point
B
through
the
relays
and
that
that
meets
that
the
requirements
would
be
good
first
set
of
tasks
to
do.
F
Okay,
thanks
I'll
leave
it
open
for
another
couple
seconds
in
case.
Somebody
else
wants
to
add
I,
see
some
plus
ones
in
the
chat
for
talking
about
protocol
talking
about
relays.
F
D
And
just
to
say,
thank
you
for
your
suhas
for
doing
the
presentation.
That's
that
was
helpful.
F
Yeah
thanks
suhas,
very
much
I
think
it's
it's
it's
good
to
see
those
things.
It's
a
little
bit
daunting
to
think
about
all
the
decisions
we
have
to
make,
but
I'm
sure
we'll
get
through
them
all
foreign.
A
I
was
just
going
to
say,
Ellen
I
think
our
next
topic
is
actually
polling
to
see
who's
going
to
be
presenting
drafts
between
now
and
and
London.
Obviously,
the
draft
deadline
is
Monday.
So
if,
if
you
haven't
got
your
fingers
to
the
keyboard,
get
them
on
now,
I
know
we're
expecting
something
from
Spencer
and
James
I
believe
we're
expecting
something
from
Luke.
Are
we
expecting
any
others?
Is
the
next
big
question.
A
B
G
Yeah
I
I
just
submit
a
dropped.
His
name
is
a
design
Choice.
The
design
is
based
on
analysis
of
the
moq
and
all
the
the
Renee
is
kind
of
the
first
first
class
citizen
of
the
mop.
But
there
are
a
lot
of
things
about.
The
Renee
is
not
very
clear.
For
example,
how
do
you
organize
the
RNA?
How
do
you
choose
the
stream
from
producer
to
Consumer
which
Renaissance
rule
if
it's
like?
The
city
is
like
a
tree
tree
topology,
but
we
have
other
choice.
G
For
example,
we
can
do
some
kind
of
overly
routing
between
between
the
renes
and
there
are
other.
There
are
other
things,
such
as
the
that
the
media
is
and
to
end
encrypted
and
the
Renee
cannot
access
a
media
or
or
Renee
can
decode
can
decrease.
The
queer
connection
I
think
I
think
the
drop
is
just
just
analysis,
so
I
would
like
to
discuss
with
people
yeah.
A
Okay,
so
please
send
us
the
the
name
of
the
draft
and
chairs
we'll
take
a
look
at
it
and
to
try
and
fit
that
into
into
the
agenda.
Maxine.
B
Yeah
hi
yeah,
my
co-author
and
I,
think
presents
an
approach
of
estimating
some
metrics
of
live
streaming,
like
Twitter
latency
on
some
packet
loss
and
related
metrics.
So
as
usual,
we
do
it
almost
the
last
date
or
tomorrow.
We
are
thinking
to
publish
it,
but
we
would
like
also
to
request
some
time
to
share
the
you
and
here,
what's
the
group
things
on
it,.
A
So
so
Maxima
I
think
that's
background
information
at
a
sufficiently
deep
level.
That
is
probably
best
if
you
send
that
to
the
working
group
mailing
list.
We're
gonna
have
a
pretty
tight
agenda
in
in
London
and
it
sounds
like
that
sort
of
Jitter
having
not
not
read.
The
draft,
of
course,
might
be
a
more
generic
interest
than
specific
to
the
design
space
from
moq.
So
if
you
could
send
it
to
the
mailing
list
now
we
can
take
a
look
at
it
as
as
a
group
and
and
go
from
there.
A
Thank
you
looks
like
we
have
Luke
and
Xavier
in
the
queue
Luke
you're
going
to
be.
The
next
presenter
can
I
take
Xavier
first.
A
G
No
that's
yeah
yeah,
we
tried
thank
you.
So,
yes,
we
we
plan
to
propose
to
publish
a
draft,
maybe
today
that's
on
on
the
type
of
relay
that
you
would
have
at
the
Ingress
point
of
g-networks
and
and
other
wireless
networks,
and
that
would
need
to
deal
with
the
congestion
and
you
know,
energy
saving
techniques.
So
then
we
need
certain
type
of
metadata.
G
So
I
think
this
can
contribute
to
the
discussion,
like
maybe
in
terms
of
of
use
case,
to
know
if
this,
this
type
of
use
case
would
be
suitable
and
also
potentially
in
terms
of
metadata.
And
you
know
what
time
of
you
know
what
time
of
what
type
of
data
data
and
where
this
meta
that
I
would
be
needed
in
the
in
the
flow.
A
A
Thank
you,
and
the
next
person
in
queue
is
actually
our
next
presenter,
so
I'm
going
to
go
ahead
and
share
your
slides.
A
H
You
go.
Thank
you
so
I
know
this
interim's
not
meant
to
be
too
technical.
So
this
is,
you
know
not
going
to
be
too
much
of
a
discussion.
This
is
going
to
really
be
covering
a
little
bit
of
the
background
of
the
previous
drafts,
even
before
the
working
group
was
formed
and
just
a
little
bit
of
progress
on
how
we're
trying
to
combine
them
come
on,
go
for
it.
So
the
number
one
goal
I
think-
and
this
really
relates
to
the
last
slides
that
suhas
have
I
think.
H
The
first
thing
we
need
to
tackle
is
how
do
we
drop
media
during
congestion
in
respect
to
the
encoding?
You
just
can't
draw
arbitrary
packets
with
media,
it's
going
to
cause
decode
errors,
but
there's
a
lot
of
different
techniques.
You
could
possibly
do
to
effectively
reduce
the
bit
rate
very.
C
H
During
congestion
and
I
wanted
to
go
over
how
various
different
protocols
to
do
that
next
slide.
H
So
the
first
thing
we
need
to
really
cover-
and
this
is
one
thing
that
it
would
be
nice
to
have
in
maybe
even
a
requirements.
Draft
I,
don't
know
this
is
like
a
background
of
media
is
how
is
Media
even
encoded?
What
are
we
dealing
with
here?
What
are
we
trying
to
put
over
the
network
so
at
the
top
I
have
a
very
simple.
This
is
what
webrtc
typically
does
or
like
really
simple,
encoders
or
even
iframe,
effectively
a
static
image
and
then
P
frames
will
reference.
H
There
are
Delta
encoded
of
a
previous
frame.
If
you
look
back
to
just
the
previous
one,
that's
the
lowest
lowest
latency.
Really,
it's
also
the
simplest.
It's
just
a
you
can
get
more
complicated.
H
You
can
start
introducing
what
are
called
B
frames,
which
can
reference
previous
or
feature
frames
and
any
number
of
them
as
well,
which
is
a
you
know,
equally
problematic,
I
love,
Hardware
encoders
will
have
a
fixed
structure
where
this
is
just
one
example,
whether
it's
ibpbp,
even
though
these
B
frames
like
frame
number
two,
is
a
b
frame.
H
It
actually
depends
on
the
third
one,
so
the
transport
needs
to
be
aware
that
frames
yeah
do
reference
out
of
order,
and
then
finally,
we
have
even
what
we
see
is
with
software
decodes
in
codes
is,
if
you
really
want
the
highest
compression
ratio,
you
pretty
much
just
have
a
spaghetti
of
references,
references
everywhere.
H
You
still
have
an
iframe,
but
it's
up
to
the
encoder.
What
references
are
made
where
this
is
very
codec
specific,
but
you
can
create
this.
This
mess
of
Tangled
mess
of
dependencies
that
need
to
somehow
be
respected
over
the
network.
H
So
one
way
of
doing
this-
and
this
is
what
meta
has
been
doing
for
a
while
now-
is
to
use
a
quick
stream
for
frame,
and
the
idea
is
there's
kind
of
this
implicit
dependency
that
this
stream
depends
on
this
stream.
So
you
should
deliver
them
in
order
during
congestion,
and
if
a
stream
is
taking
too
long
and
it's
not
high
priority,
you
can
send
a
reset
stream
frame
to
effectively
drop
it
during
congestion.
H
This
is
kind
of
similar
to
how
webrtc
Works,
actually
in
RTP
in
general,
is
kind
of
package
at
a
frame
level,
but
there's
no
explicit
information
on
the
wire
about
these
dependencies.
That's
why
those
dotted
arrows
everywhere
next
slide.
H
H
We
just
delivered
them
in
an
order
that
the
decoder
can
handle
just
fine
and
there's
an
ex
there's
an
idea
of
prioritization
such
that
newer
content
will
arrive
for
older
content
and
that's
how
you
can
effectively
starve
and
drop
during
congestion,
so
yeah
so
Russia's
previous
draft
out
there
warp
is
another
draft,
that's
being
published
next
slide.
H
And
the
the
third
notable
draft
that
we're
trying
to
combine
is
a
quick
R.
It's
a
concept
of
it's
far
more
complicated
than
this.
By
the
way,
it's
a
effectively
a
pub
sub
network
with
the
important
distinction
that
there's
a
header
in
the
front
of
each
stream
that
informs
a
relay
on
how
it
should
that
that
stream
should
be
delivered,
such
as
identification,
prioritization,
timestamp.
H
Any
information
required
for
a
a
larger
fan
out
decision
under
the
hood.
It
uses
rush
or
warp.
There
is
a
section
about
datagrams
as
well,
but
just
to
simplify.
It
uses
one
of
the
two
under
the
hood
next
slide.
H
So
my
goal
is
being
to
try
and
combine
the
parts
of
the
draft
that
all
agree
with
each
other
to
try
and
find
a
single
approach
that
can
do
either
Style
at
least
either
rush
or
warp
style
of
delivery,
and
this
I'll
have
this
ready
by
Monday
I'll
at
least
have
a
draft
published,
but
just
a
little
teaser.
The
idea
is
that
you
can
have
a
stream
for
any
number
of
frames
is
effectively
the
compromise.
H
So
you
can,
you
can
do
Rush
where
you
can
have
a
stream
per
frame,
and
you
can
do
warp
when
you
can
have
multiple
frames
per
stream
and
then
there's
a
middle
ground
where
you
can
just
pick
and
choose
whatever
frames
are
on
the
stream.
The
important
caveat,
though,
is
that
there
is
a
header
within
each
stream
that
says
that
this
stream
depends
on
the
stream
or
like
some
information
for
both
relays
and
also
to
know
what
streams
should
be
delivered
in
what
order
to
best
respond
to
congestion.
H
Next
slide
so
yeah,
like
I,
said
London
draft
I'd
love
to
talk
more
about
the
technical
details,
I,
don't
wanna
I,
don't
have
the
time
to
to
talk
too
much
of
this
interim
and
the
big
thing
I
really
want
to
plant
in
people's
head
is
kind
of
alluding
to
what
I
think
a
sort
of
chat
just
earlier
is
how
will
CDN
support
this
in
general?
I?
Think
that's
an
ongoing!
That's
just
something!
I
I
just
want
people
to
start
thinking
about
it.
H
Do
you
want
a
pub
sub
model
or
are
we?
Is
there
some
way
we
can
map
these
streams
to
a
traditional,
pull-based,
CDN
yeah
and
that's
not
yeah
I.
Think
that's.
Probably
one
of
the
next
bigger
things
to
to
identify
is,
if
you're,
trying
to
combine
all
the
use
cases
in
one
protocol
next
slide
and
yeah.
A
lot
of
people
have
helped
with
this
draft.
H
Quite
a
few
names
up
here,
but
the
main
authors,
I
would
say
Luke
for
for
warp,
Sue
Has
for
a
quick,
R
and
kiril
for
rush,
but
there's
been
a
collaborative
effort
amongst
a
lot
of
people.
So
thank
you.
A
Very
much
and
just
for
the
folks
here,
I
think
we.
We
are
counting
this
as
a
preview
of
London.
Rather
than
opening
this
for
discussion
today,
we
only
have
like
six
minutes
left,
so
definitely
not
enough
time
to
dive
into
the
details
there.
So
here
it
looks
like
you're
in
queue.
Is
it
to
dive
into
the
details,
because
if
so
I'm
going
to
ask
you
to
hold
your
water
until
the
draft's
out
and
then
take
to
the
list.
A
It's
Spencer
also.
D
A
That
is
very
encouraging
and
we
really
appreciate
the
work
on
that.
I
will
encourage
you,
as
you
put
the
draft
together
to
think
about
what
folks
you
need
to
reach
out
to
whether
they
be
other
folks
who
do
similar
things
like
the
Youtube
crew
or
the
CDN
folks,
or
any
of
our
video
Codec
Posse
here
at
the
ITF,
because
we
may
want
to
get
early
review
from
quite
a
variety
of
people
if
this
is
going
to
be
kind
of
the
the
basis
of
ongoing
work
here.
A
I
have
seen
in
in
the
past
a
couple
of
situations
where
we
didn't
do
that
early
review
quite
broadly
enough
and
ran
into
later
problems.
So
after
the
draft
is
out,
expect
it
to
be
passed
from
hand
to
hand
pretty
rapidly.
B
A
H
Replied
to
my
thread
on
the
mailing
list,
with
a
link
to
the
the
current
draft
full
of
twos
and
and
the
GitHub.
If
anybody
would
like
to
take
a
look
and
leave
feedback.
Okay,.
A
That
would
definitely
be
great,
so
I
think
that
brings
us
to
the
end
of
our
agenda.
I
want
to
thank
the
note
takers,
especially
I,
really
appreciate
it
for
work
they
put
into
to
grab
some
notes
for
us.
Is
there
anything
else
we
need
to
cover
today
before
we
adjourn
and
start
getting
everybody
ready
for
their
final
run
at
drafts
for
London.
A
Okay,
thanks
again,
and
we
will
see
you
on
the
mailing
list.