►
From YouTube: 2021-10-06 Multicast W3C Community Group Meeting
Description
Discussed AMT implementations and their v4/v6 support status, potential IETF hackathon project feeding the multicast payloads to webcodecs, TPAC organizing. Minutes here: https://github.com/w3c/multicast-cg/blob/main/meetings/minutes/minutes-2021-10-06.md
A
Great,
we
are
recording,
thank
you
all
for
coming,
oh,
should
I
I
can.
I
can
stop
this
and
restart
it
if
you
need
to
erase
anything.
A
All
good,
okay,
great
great
thanks
everyone
for
coming.
I
guess
if
no
one
has
an
agenda
bash
to
the
did.
Everyone
get
a
chance
to
read
the
email
from
last
night.
B
A
A
All
right,
so
I
guess:
let's
go
over
the
teapot,
that's
that
so,
first
off
our
scheduled
slot
for
the
multicast
community
group,
that'll,
be
it
I
put
a
link
in
there.
It's
I'm
hoping
to
get
a
few
new
people
to
be
interested
in
joining.
A
Relatively
likely,
but
I'm
not
sure
exactly
what
to
expect
so,
maybe
maybe
we'll
have
a
few
new
people
coming
in
with
some
interest.
A
A
Unfortunately,
I
have
not
been
able
to
get
to
that
since
we
discussed
it
two
months
ago,
but
I
did
put
up
that
demo
video
and
do
some
of
the
other
things
I
had
hoped
to
do
so,
what
we'll?
So,
what
we'll
probably
do,
or
what
I'll
plan
for
in
the
in
the
interest
of
having
new
people
around
will
be
a
little
bit
of
intro
and
then
get
into
you
know
what
the
what
the
work
will
be
ongoing
that
will
that
we'll
need
to
solve
before
this
can
be.
A
You
know
integrated
into
a
browser
and
rolled
out
I'll,
probably
try
to
to
give
an
overview
of
the
security
situation
and
what
would
be
a
little
bit
of
the
useful
work
there
and
I
guess
yeah
any
other
suggestions
that
I
should
fold
in
there.
A
Anyone
want
to
take
some
of
the
speaking
time
anything
you
want
to
present,
perhaps
or
or
talk
about.
A
B
That's
good,
I
think
my
only
question
would
be
like
other
particular
people
that
we
want
to
invite
specifically
to
join.
Or
is
I
mean?
I
I
think
you
will
get.
You
know
a
number
of
people
who
are
curious.
You
know,
tpack,
being
you
know,
kind
of
open
to
lots
of
you
know
all
the
w3c
members,
so
you
may
well,
you
may
well
get
you
know,
people
who
have
maybe
not
familiar
so
far
with
what
we're
doing
and
are
curious.
B
So
I
think
yeah
having
that
introductory
material
is
is
is
good,
but
I'm
also
thinking
yeah
whether
there
are
other
people
that
we
should
specifically
try
to
bring
along.
A
Yeah,
that's
a
good
question.
I
guess
I've
been
thinking
in
terms
of
categories
of
people.
Some
of
the
individual
people
I
I
would
love
to
have,
I
know,
are
super
busy,
but
you
know
here,
I'm
thinking
of
so
like
harold.
Alvestrand
has
been
supportive
of
this
work
before
and
you
know
obviously
has
hugely
important
experience
in
in
with
the
webrtc
work
and.
A
You
know
just
navigating
the
standards
bodies
in
general,
there
were
a
few
people
at
the
at
the
last
ietf
barb
off
that
seemed
potentially
interested,
but
I'm
not
sure
how
much
overlap
there
was
with
w3c
jason
folks.
I
guess.
A
Sort
of
what
working
working
you
know,
people
who
will
tell
me
what
I'm
doing
wrong
in
the
browser,
implementations
and
stuff
and
who
will
dig
in
on
that
and
maybe
help
correct
it.
Even
you
know,
I
think
I
got
a
little
bit
of
feedback
on
that,
but
I
have
not
put
as
much
work
into
that
since
kind
of
kicking
it
off
that
that
I
would
like
to
as
what
I
would
like
to
have
done.
A
I
would
like
to
end
up
with
with
more
progress
on
that
front,
but
then
there's
also
just
the
getting
the
basics
of
the
networking
working.
B
A
Maybe
some
googlers
like
if
there's
people
at
google
who
can
tell
me
where
I'm
like
gonna
bump
into
a
brick
wall
on
on
the
on
the
chromium
side
or
alternately
like
someone
embedded
in
firefox
more
that
would
that
would
move
the
whole
project
over
or
something
that
would
be
interested
in
moving
the
whole
project
over
and
starting
to
help
with
that.
You
know,
I
don't
really
care
where
it
gets
started,
but
you
know
some
kind
of
a
like
mentor
in
the
in
the
browser
world.
I
think,
would
be
helpful.
A
C
We
got
at
the
moment,
there's
basically,
is
you
and
another
guy
jack
akamai
working
on
this
in
terms
of
development
working
on.
A
C
Have
said
this
is
this
looks
interesting
and
we'd
like
to
keep
an
eye
on
it,
but
not
too
many
people
leaping
forward
to
open
a
terminal,
editor
and
start
hacking
away.
A
Yeah,
I
did,
I
did
have
a
group
briefly,
but
they
got
reassigned.
I'm
still
working
on
getting
another
group
we'll
see
if
that
pans
out.
We
should.
I
should
find
that
out
this
month,
but.
C
In
terms
of
feedback
and
so
on
from
other
people
outside
your
organization,
are
you
getting
much
of
that
at
the
moment?
Is
that
that's
really?
What
we're
talking
about,
I
guess,
is
you've
not
got
a
lot
of
feedback
from
other
iatf
and
w3c
people.
A
Yeah
from
ietf
I
got
a
bit
I
I
have
a
recording
online
of
the
barbot.
There
were
some
20
to
maybe
almost
30.
A
I
think
I
want
to
say
26
ish
participants
at
at
a
barb
off
meeting
I
had
at
the
lat
ietf
people
seemed
yeah
supportive,
but
not
going
in
and
saying
hey
I
want
to.
I
want
to
write
some
of
the
code
on
this,
which
I
don't
see
a
whole
lot
of
yeah
yeah
yeah.
I.
C
Wondered
if
that
was
how
this
was
gonna
roll,
that
a
lot
of
people
would
turn
up
for
the
initial
thing
to
see
hey
what
are
they
doing
and
then
fade
away.
Unfortunately,
I
faded
away
myself
a
little
bit
there.
I'm
afraid
I've
not
been
terribly
well
for
some
time,
so
I'm
quite
enthusiastic,
but
I've
got
to
manage
my
own
energy
levels.
C
I
unfortunately
contracted
covered
last
year
and
yeah.
What
I
didn't
realize
is
sometimes
it
it
takes
quite
a
long
time
to
recover,
and
in
this
case
18
months,
I'm
still
not
quite
back
on
my
feet,
but
I
seem
to
be
getting
there
so,
and
I've
got
my
own
project,
which
has
a
lot
of
catching
up
to
do
at
the
moment,
so
sure.
C
C
Ipv6
multicast
and
transitional
tunneling
and
things
so
that
I
think
got
your
interest
before
was
that
I
had
a
little
chat
app
that
I'd
built
that
used
websockets
and
things
and
multicast
in
the
background-
and
maybe
maybe
there's
something
we
can
do
with
that.
If
it
can
do
native
multicast,
then
it
can
throw
with
the
websocket
and
start
talking
directly
to
the
network,
which
would
be
nice.
A
Welcome,
I
see
we
have
a
new
participant.
I
just
wanted
to
admit.
I'm
not
sure
how
to
pronounce
your
name
is
that
michiel.
A
Great
thank
you
for
joining.
I
just
wanted
to
make
sure
you're
aware
we're
recording
if
we,
if
you
have
anything
you'd
like
redacted,
then
I'll
cut
it
out
afterwards,
no.
A
So
just
wanted
to
make
sure
that's
on
your
radar.
Thank
you
for
joining
and
and
welcome,
and
let's
see
so
back
to
so
ipv6.
This
is
tim.
Chon
is
the
first
one
who
asked
for
this
and
that
that
I
would
say
was
just
within
the
last
two
weeks
we
were
talking
about
so
they're
standing
up
some
amt
relays
and
some
beer
forwarding
capabilities
in
in
the
giant
network.
A
That
was
a
little
bit
later
in
the
in
the
email.
I
had
a
a
thread
to
the
link
to
the
discussion
thread
about
the
about
an
ietf
hackathon
upcoming.
I
believe
that
would
be.
I
forget
exactly
when
that
is
early
november.
I
think
possibly
the
week
before
with
some
overlap.
A
As
I
understand
it,
the
giant
implementation,
which
is
a
new
implementation,
was
written
to
support
ip4
and
six.
The
the
amt
relay
and
gateway
work
in
the
so
there's
two
places.
There's
the
in
in
my
grandfield
troll
repos
in
github,
there's
amt
relay
d
and
and
the
amt
gateway
those
both
I
did
put
it
in
ip6
or
support
a
couple
years
back
and
you
know
ran
it
once
in
lab
and
it
it
worked
fine-ish.
A
So
I
I
think
that's
the
status
on
that.
I
don't
have
any
actual
relays
running.
Yet
it's
one
of
the
the
things
that's
been
on
my
list.
I
would
love
to
get
one
up
for
I
I
was
hoping
to
get
one
up
for
tim
and
and
you
to
send
just
send
some
of
that
same
kind
of
vlc
traffic
that
you
can,
if
you
can
get
it
shipped
to
your.
A
To
your
end
device
by
using
these
the
gateway,
then
you
would
then
you
in
theory
you
should
be
able
to
watch
it
over
watch
it
on
vlc
and
it's
a
useful
test
to
play
with
when
you're.
You
know
trying
to
trying
to
work
on
something
and
then
the
the
browser
work
that
we're
proposing
for
this
next
step
would
be
to
also
try
to
render
that
I'm
not
sure
how
that's
gonna
go,
but
but
in
theory
like
whatever
is
it
vlc
is
doing
to
play.
A
That
should
be
pretty
similar
to
what
webcodex
is
doing
and
I'm
not
sure
if
it'll
be
pickier
on
the
codex.
If
there's
like
malformatted
things
or
or
stuff,
not,
you
know
missing
packets
or
anything.
A
If
it
is,
then
it
it
might
not
play
very
well,
but
if
it's,
if
it's
not,
then
maybe
we
can
get.
You
know
some
frames
rendered
kind
of
what
we're
aiming
to
do
in
this
in
this
upcoming
hackathon.
I
think,
and
I've
also
got
I've
also
got
the
sort
of
the
the
stuff
I
showed
in
the
demo.
Video
which
is
based
on
our
akamai
has
a
product.
A
That's
used
as
a
tv
service
inside
a
wall
garden,
multicast
deployment
in
isps
and
we
ported
that
into
webassembly
the
receiver
side
of
that
into
webassembly,
and
that
was
that
was
what
was
running
on
top
of
the
on
top
of
the
browser
for
that
demo.
Video.
If
you
got
a
chance
to
see
it,
and
so
that's
a
reliable
transport,
that's
doing
it's
like
rebuilding
the
segments
out
of
out
of
the
data
that
it
got
over
over
multicast
and
then
feeding
in
segments
that
were
that
were
transferred.
A
I
think
they
are
checks
unchecked
before
feeding
them
in
and
then
there's
a
unicast
fallback
if
it
needed
if
it
needed
that.
So
that's
what
was
running
for
the
for
that
piece.
A
But
which
is
why
we're
doing
the
vlc?
Basically
for
now
at
least-
and
I
think,
when
we
get
to
the
to
the
quick
implementation,
then
that
will
end
up
being
kind
of
the
same
way
where
inside
inside
the
quick
library.
As
I
understand
it,
you're
going
to
be
sending
segments
there's
a
sort
of
reliable
transport
that
will
rebuild
the
segment
on
the
other
side.
And
then
you
feed
the
segment
into
into
mse.
A
I
think
to
play
that
so
that's
and
in
theory
that
might
be
possible
to
make
it
like
inside
the
xhr
instead
of
inside
the
instead
of
a
web
transport
like
api,
where
we
got
a
read
stream,
a
readable
stream
that
we're
reading
you
know
all
the
packet
payloads
out
of
yeah,
but
that
work
is
still
that'll,
be
a
relatively
different
path
inside
the
inside
the
implementation.
A
So
we'll
see
we'll
see
what
that
looks
like
and
I
I
would
like
to
get
it
so
that
datagrams
would
come
through
a
similar
kind
of
readable
stream
path
that
could
be
used
in
the
same
way
in
the
quick
implementation
as
well.
So
I
don't
know
how
feasible
that
will
be,
but,
as
I
understood
it
from
the
last
discussion,
that's
not
there
yet
in
the
nghq.
B
Yeah,
there's
no
supporting
that
yet.
A
Yeah,
okay,
anyway,
I
forgot
what
your
question
is:
did
we
cover
it.
C
Yeah,
essentially
that
it's
it's
mainly
might
be
the
full
support
at
the
moment.
This
is
the
original
question,
but
that's
fine.
This
is
all
good
for.
C
A
Get
if
I
managed
to
get
a
v6
relay
up
in
the
next
you
know
in
the
next
week.
I
anticipate
there
might
be
a
little
bit
of
like
my
launching
scripts
for
for
bringing
up
a
sort
of
vlc
compatible
thing.
I'm
not
sure
they'll
work
with
ip6
we'll
have
to
see
actually
I'm
pretty
confident
they
don't,
but
it
might
not
be
hard
to
add
them,
because
I
think
that
aws
instances
will
have
ip6
by
default
or
at
least
allow
you
to
turn
it
on
easily.
A
But
yeah
now
that,
now
that
a
couple
more
people
are
looking
at
it,
then
it's
the
second
request.
I've
had
to
actually
run
it
that
way,
so
so
I'd
like
to
get
I'd
like
to
get
that
up.
C
Our
approach
with
with
laborcast
is
basically
v6
only
and
the
idea
is
given
that
multicast
is
not
that
widely
supported
that
we'd
need
to
tunnel
to
get
multicast
working.
Then
you
may
as
well
tunnel
and
get
v6
multicast
as
opposed
to
just
just
v4.
So
everything
we
do
is
built
around
that
and
if
there
was
only
v4
multicast
available,
then
we'd
still
tunnel
just
to
get
v6
going
yeah.
It's
just
some
of
the
things
we're
doing
require
v6
only
multicast
things
like
the
groups
and
so
on.
C
A
In
in
ssm.
C
A
C
A
I
think
that's
the
latest,
but
yeah,
if
you,
if
you
have
a
another
one,
you
know
point
me
to
that.
To
that
reference.
A
And
okay,
so,
oh
and
and
we're
doing
ssm,
because
asm
has
been
deprecated
for
interdomain,
so
I'm
restricting
any
of
my
interest
to
ssm
multicast.
Just
for
I
can
get
you
that
reference
too,
but
that's
one
of
the
more
recent.
C
A
Yeah,
it
doesn't
the
if
you,
if
you
ever
want
to
read
the
sad
story,
take
a
look
at
rfc
4611
for
msdp
the
I've
never
seen
such
a
discouraged.
Looking
intro
section
talked
about
how
you
couldn't
couldn't
reach
consensus,
et
cetera.
It
was
bad,
hey,
kyle,.
E
E
A
No
worries,
let's
see
we
were
just
about
to
get
back
on
track.
We
had
a
a
quick
discussion
about
ip6,
actually
that
that's
part
of
bringing
us
on
track
overall,
I
think
so
all
good,
let's
see
but
I'll
I'll,
send
out
an
update.
If
I
bring
up
a
v6
relay
and
some
v6
traffic-
or
maybe
I
should
say,
or
some
v6
traffic-
oh
it
would
you
would
you
use
the
v4
tunnel
if
you
had
a
v4
tunnel
for
v6
traffic
or
like?
C
Well,
the
multicast
we
use
is
ipv6
only
so
if,
if
ipv4
multicast
is
available,
we
wouldn't
use
it
with
tunnel.
That's
all
I
was
saying
there
so
ipv4
unicast
is
sufficient
because
we
just
use
it
to
tunnel
and
get
to
some
real
multicast.
A
Think
so
I'm
not
sure
how
you
would
tunnel
it.
If
the
group
address
that's
being
sent,
you
could
use
m.
We
actually
do
some
of
that,
although
this
one
also,
I
don't
think
supports-
maybe
we'll
take
that
offline.
It's
a
little
bit
sideways
for
the
for
the
web
development
stuff,
but
yeah.
We
do
want
v6
to
work
in
the
browser,
so
in
theory
it
should.
I
would
love
to
try
it
out.
A
All
right
so,
let's
see
I
don't
know
whether
we'll
get
the
slotted
web
transport.
I
also
just
found
out
this
morning
the
slot
in
a
second
screen
they've.
They
put
me
down
for
an
hour.
I'm
not
sure
why,
but.
A
The
overlap
with
second
screen-
I
wasn't
too
sure
about
because
I'm
not
sure
how
applicable
multicast
is
gonna,
be
for
the
for
the
second
screen
use
cases.
As
far
as
I
could
see,
that's.
B
You
know
launching
video
streams
or
launching
websites
on
a
second
screen
device,
so
you
can
and
it's
largely
designed
around
discovery
of
devices
on
your
local
network,
but
there
is
a
protocol
for
streaming
media
between
devices
so
yeah.
I
think
I
think
it's
great
that
you've
yeah
that
you've
got
their
attention.
That's
that
sounds
really
positive,
oh
yeah,
but
I
was
kind
of
wondering
what
kind
of
more
specifically
what
the
potential
collaboration
with
them
might
be.
A
Yeah,
so
the
the
first
potential
collaboration,
the
one
that
was
most
interesting
to
me.
Actually
I
put
it
in
the
in
the
issues
list
for
the
link
to
that.
A
A
Let's
see
so
my
my
first
interest
is
the.
A
Is
the
privacy
issues
because
I
think,
there's
a
so
there's?
Maybe
two
privacy
footprint
exposure
issues
for
for
this
group,
and
one
of
them
is
shared
with
second
screen,
which
is
about
a
privacy
exposure
footprint
to
the
local
network
that
is
controlled
by
a
website
that
you
visit
right.
So
you
go
there.
A
It
uses
your
apis
to
do
something
on
your
local
network,
and
you
know
I'm
not
sure
if
there's
a
if
there's
a
threat
model
that
comes
from
exposing
information
to
your
local
network,
that's
controlled
by
the
website,
and
so
the
the
I
wanted
to
pick
their
brains
a
bit
about
whether
they've
encountered
any
abuses
with
their
api
and
whether
they've
examined
any
of
the
threat
models
there.
A
A
So
this
exposure
footprint
is
an
area
of
concern
and
and
there's
questions
about
how
abusable
it
it
might
be.
The
other.
The
other
related
issue
for
multicast
in
general,
especially
for
the
sort
of
not
just
within
the
lan
multicast,
that
you
use
for
discovery,
would
be
that
the
exposure
of
a
join
to
the
to
the
propagating
network
is
also
necessary.
In
some
sense,
the
network
will
know
that
it's
forwarding
traffic
at
the
last
hop
router.
A
It
will
know
where
it's
forwarding
subscribed
streams
and
there's
another
separate
issue
about
exposure
footprint
for
there.
That
is
not
really
something
that
the
second
screen
group
would
have
hit,
but
exposure
from
discovery
protocols
within
the
network-
that's
that's
controlled
by
the
website,
is
one
of
the
one
of
the
sort
of
open
issues
for
whether
it's
going
to
be
a
blocker
and,
if
so,
of
what
form
or
whether
it
would
just
sort
of
add
constraints
into
what
can
be
done
in
the
multicast
group.
C
Didn't
apple
have
some,
I
think
in
the
last
month
or
so
to
do
with
blocking
multicast
in
their
app
store,
due
to
being
able
to
use
it
to
fingerprint
the
local
network.
Yeah.
A
So
it
it
wasn't.
Yes,
good
good
point,
I
I
I
looked
at
that.
I
wasn't
sure
what
to
think
of
it
at
first,
but
I
think
it
actually
might
be
helpful
towards
this
towards
this
concern,
because
it
means
that,
at
least
on
that
platform,
those
platforms,
the
developer,
you
know,
there's
a
mechanism
by
which
to
shut
it
off.
If
it
gets
abused,
you
know.
A
So
the
idea
is
that
you
know
well-behaved
developers
will
register
their
usage
and
what
it's
for
and
then,
if
they're
caught
misbehaving
in
some
way,
then
then
you
know
it's
not
just
that.
There's
no
answer
they
can
their
app
can
be
sort
of
shut
off
and
and
not
have
that
capability
and
prevented
from
using
that
capability
anymore.
A
So
I'm
not
sure
whether
that's
a
thing
that
would
ever
happen
on
some
of
the
other
platforms.
You're
sort
of
you
know
I
I
I
don't
know
what
it
would
do
on
windows
to
make
that
happen,
for
example,
or
linux.
A
Or
or
android
even
but
but
yeah,
it's
it's
an
interesting
take
on
things,
and
I
think
it
tends
to
help
to
help
concern
address
concerns
about
that
problem.
Although
I
don't
think
it's
necessarily
a
full
solution.
A
But
you
might,
you
might
see
similar
things
at
the
os
layer
where
untrusted
apps
could
be
blocked
at
the
you
know
by
the
os
if
they
haven't
gone
and
sort
of
requested.
Permissions
in
some
form
expect.
B
Yeah,
so
that
that
sounds
really
positive.
I
know
that
in
the
second
screen
group
there
there
has
been
a
quite
a
bit
of
thought
put
into
the
protocol
design
and
the
layering
in
terms
of
what
what
information
gets
gets
exposed
at
what
what
level
and
the
establishment
of
the
security
connection
you
know,
and
so
so
I
think
yeah.
I
think
you
should
be
able
to
have
a
useful
discussion
actually
about
the
I
mean.
Obviously,
the
details
will
differ
but
like
that,
the
principles
that
they're
applying
and
could
well
be
helpful.
A
A
Excuse
me
not
used
to
talking
that
much.
I
can
imagine
that
there's
a
potentially
someone
is
interested
in
in
using
multicast
like
subscribing
to
the
same
multicast
stream
from
two
different
screens
so
that
you
can
or
multiple
different
screens
so
that
you
can
have
a
sort
of
synchronized
playback
that
doesn't
have
to
necessarily
traverse
the
local
network.
A
A
If
anybody
does
have
that
in
mind,
then
I
don't
know
what
strategy
will
be
for
handling
the
sort
of
wi-fi
constraints
and
whether
that
will
make
it
sort
of
a
non-starter
in
the
end,
but
that's
kind
of
my
expectation
without
unless
there's
a
sort
of
credible
way
to
to
work
with
that.
A
A
You
could
imagine
like
in
a
well.
I
guess,
like
in
a
bar
environment,
they'll,
usually
be
networked
right
or
they'll
usually
be
wired.
I
mean
and
wired
you
wouldn't.
You
would
potentially
be
able
to
do
that,
but
you
can.
You
can
imagine,
like
50
screens
all
showing
the
same
thing
they
could
be.
They
could
be
maybe
usefully
subscribed
to
the
same
multicast
stream,
even
in
f5,
because
it
would
use
use
less
airtime
than
trying
to
do
it.
Unicast
from
some
from
some
central
location
or
between
each
other
peer-to-peer.
A
D
I
think,
in
that
case
they
would
also
have
to
be
able
to
publish
right,
because
I
think
that
they're
also
publishing
from
the
browser
to
other
devices
is
that
something
that
you're
considering
right
now
or
is.
It
is
only
the
read
side
of
things
for
the
time
being.
A
Yeah
for
multicast
it's
going
to
be
hard
enough
to
do
receive
only
the
the
proposed
api
is
receive
only
now
it.
You
know,
I
don't
want
to
put
a
nail
in
that
coffin
of
it.
It
might
be
possible
to
coordinate
two
screen.
Two
streams
to
you
know
both
subscribe
to
the
same
multicast
stream
from
from
outside
and
for
that
to
end
up
being
worthwhile,
but
I
think
there's
gonna
have
to
be
some
like.
A
You
probably
won't
get
gains
if
that's
happening
over
wi-fi
is
my
only
it's.
My
only
caveat
there.
So
if
it's,
if
it's
happening
in
in
a
wired
network
or
potentially
in
a
cell
network,
then
that
might
be
that
might
end
up
being
worthwhile,
but
on
a
wi-fi
network,
there's,
unless
maybe
wi-fi
seven
does
a
better
job
with
multicast,
it
might
end
up
being
ugly.
A
A
You
know
this
is.
This
is
like
why
udp
sockets
were
were
turned
off.
You
know
you
can
use
them
to
to
drive
a
dos
attack
or
something
so
you
can't
just
give
control
over
web
pages.
Give
control
to
web
pages
to
you
know
to
send
out.
A
Yeah
so
we're
looking
at
the
multicast
apis
as
as
receive
only
probably
forever
there's
also
authentication
stuff.
That's
that's
hard
to
do.
Otherwise,
although
I
guess
the
authentication
is
probably
equally
hard
for
a
four
second
screen,
but
limiting
limiting
it
to
the
to
the
land,
you
know
might
help
with
that.
E
So
I
can
imagine
a
case
in
the
future
where
somebody
has
a
brilliant
idea
to
use
multicast
to
do
to
do.
E
Video
conferencing
to
you
know
like
a
direct
user
to
user
sort
of
thing,
but
you'd
be
sending
to
a
multicast
group
and
if
no
one
subscribed
to
it,
then
the
packets
aren't
going
to
get
very
far
so
may
I
could
maybe
see
a
case
for
for
sending
allowing
udp
packets
to
be
sent
to
a
group
to
multicast
group
address
only,
but
even
then
there
would
have
to
be
some
sort
of
protection
against
you
know
using
it
to
send
to
some.
E
E
No,
I
mean
I
mean
I
I
guess
what
I'm
saying
is
that
you
don't
you
don't
want
somebody's
browser
to
be
able
to
spoof,
you
know
to
to
be
able
to
inject
packets
into
somebody
else's.
A
I
don't
think
that
would
yeah.
That
would
be.
That
would
be
beyond
the
browser's
capability.
I
think
I
would.
E
A
E
A
A
E
A
Yeah
I'll
say
that's
probably
out
of
scope
for
for
that,
for
this
group
you
know,
and
some
of
the
expertise
may
carry
over
yeah
in
the
future,
but
yeah
yeah
specialized
things
to
do
discovery
of
local
things.
You
know
this.
This
makes
some
sense
and
it
uses
outbound
multicast.
But
that's
not
that's
not
what
we're
doing
here.
A
Right
did
I
have
any
takeaways
from
that?
I
guess
any
yeah.
Anybody
have
anything
to
add
to
to
those
two
insights
say
anything.
I'm
missing.
C
Well
I'll,
take
carl's
carl's
point
that
udp
just
allowing
outbound
udp
unicast
is
obviously
quite
a
risk,
but
outbound
udp
multicast
is
not
a
similar
risk
at
all
as
it's,
it
is
going
to
be
dropped
by
the
switch
or
the
first
top
router.
C
There's
no
way
you
can
do
an
amplification
attack
using
multicast.
In
that
way,
the
two
are
just
just
not
similar.
I
think
we
might
be
protecting
against
a
threat.
That's
not
there
in
the
same
way
that
the
ttl
for
multicast
is
by
default,
set
one
when
there's
really
no
reason
for
it
to
be.
If
there
was
any
risk
you
know,
I
would
love
there
to
have
been
a
risk.
C
It
would
mean
that
multicast
worked
on
the
internet
and
we
wouldn't
be
sitting
here
now
having
this
meeting
if
ttl1
was
necessary,
we
wouldn't
all
be
here
now,
but
it
found
its
way
into
an
rfc,
because
somebody
was
thinking
unicast
when
they
wrote
it,
and
I
think
we're
in
danger
of
doing
the
same
thing
here
by
saying
that
this
would
be
read-only.
C
C
E
The
that's
exactly
the
use
case
I
was
thinking
of,
is
you
know,
instead
of
instead
of
having
a
you
know
hub
and
spoke
model
like
you
know,
webex,
where
you
know
you
send
video
to
the
hub
and
then
it
sends
it
to
everybody.
You
just
send
your
stream
to
everyone
who's.
Watching
I
mean
I
could
certainly
like
that.
That
seems
like
a
perfectly
natural
use
case.
D
There
is
some
problems
with
that,
though,
because
it
opens
up
a
whole
bunch
of
things
like
ad
networks
could
start
creating
shadow
networks
on
your
local
end
to
fingerprint
your
network
and
that
sort
of
thing,
so
it
gets
very
hairy
very
quickly.
D
I've
seen
a
whole
bunch
of
you
know
people
trying
to
do
initiatives
in
the
past
with
this
sort
of
thing
that
all
failed.
So
I
think
jake's,
probably
right
in
keeping
it
to
read
only
for
the
time
being,
just
to
keep
out
of
that
whole
area
of
concerns.
E
A
Fair
point
it
I
I
shouldn't
say
never
things.
Many
things
can
happen
in
the
future,
but
yeah
that
is
not
our
present
focus.
I'll
just
say.
D
A
D
I'll
just
very
quickly
mention
that
I
tried
to
generalize
what
the
second
screen
group
has
been
doing
and
tried
to
bring
their
protocols
into
a
generic
web
api,
and
I
made
a
small
suggestion
for
that
on
the
community
group.
What
is
it
called
again.
A
Yeah
I'd
be
interested
to
see
that
thread.
Certainly
this
is
this
sounds
like
the
kind
of
thing
I
want
to.
I
want
to
see
where
the
bombs
are,
so,
if
you
could
at
least
send
that
to
me,
maybe
send
it
to
the
public
multicast
w3
list.
If
you
could.
A
A
Any
I
guess
interest
in
the
hackathon.
Perhaps
this
was,
I
was
pleased
to
see
that
lenny
landed
on.
You
know
one
of
the
things
that
was
on
our
list
from
from
previous
meetings
like
we
could
try
receiving
the
dlc
stuff
in
and
passing
it
into
web
codex
and
see
what
happens.
So.
It
looks
like
we'll
get
a
chance
to
try
that.
E
Okay,
what
is
the?
What
are
the
since?
I
haven't
been
doing
any
hackathon
stuff
since
since
the
before
times,
what
is
the
the
structure
of
them
now.
A
You
know
you
you
it's
similar
but
remote,
okay,
so
the
it'll,
be
you
know
you
you,
the
champion
registers
it
in
the
wiki
page
and
and
sends
a
notice
to
hackathon
itf.org.
A
You
know
you
get
together
with
your
group.
I.
A
I
think
they
move,
I
think
they
made
it
the
week
before
so
that
you
know
people
are
just
gonna
sort
of
and
within
the
group
you
can
sort
of
structure
it.
However,
you
want
at
the
end
of
the
at
the
end
of
the
meeting.
I
think
I
think
they
have
a
similar
sort
of
you
know
three-minute
lightning
talks
to
tell
what
your
group
did
and
how
it
went
that
people
can
participate
in
so.
A
Yeah,
so
basically,
you
know
no
rules
about
what
you
work
on
no
requirement
to
present
you
just
sort
of
say
that
you're
going
to
be
working
on
stuff
and
hope
that
you'll
get
an
extra
person
or
two
besides
the
people
that
were
already
on
the
project.
You
know.
A
So
yeah
anyway,
we'll
we'll
be
working
on
that
I'll,
send
an
update
if
we
get
anything
or
if
we
find
anything
out.
But
I'm
hopeful
that'll
be
a
little
bit
useful.
It's
not
the
direction
that
that
will
want
to
be
with
the
quick
work,
but
it
might
be
a
relevant
use
case
for
for
the
datagrams.
A
If
we
can
add
that
in
afterwards,
like
the
vision
I've
got,
there
is
that
you
can
have
existing
udp
multicast,
capable
things
like
vlc
and-
and
you
can
just
stick
something
in
the
front
where
you're
sending
stuff
and
have
it
sort
of
encode.
A
Whatever
udp
packets
were
sent
and
as
long
as
they're,
you
know
within
the
right
constraints.
Maybe
maybe
you
can't
do
it
over
a
certain
size
or
something?
That's
that's
a
little
bit
smaller
than
your.
You
know
normal
udp
packet
size,
but
then
outside
that
you
can
turn
that
into
a
quick,
udp
datagram
and
you
know,
build
the
authentication
hash
and
send
that,
through
through
your
ambi
kind
of
stream,
and
and
get
it
delivered
to
a
browser,
and
then
unpacked
and
and
handled
in
whatever
way
is
appropriate
for
that.
A
For
that
data
stream
and
from
the
browser
point
of
view,
it
will
look
the
same
as
the
current
api
that
that
just
sort
of
takes
a
raw,
udp,
payload
and
hands
it
off
to
the
to
the
api,
but
it
will
have
the
the
quick
part
built
around
it.
So
I
think
I
think
applications
that
do
this
will
have
the
same
kind
of
workflow
and
it
it
opens
the
door
to
anything
that
already
existed
as
long
as
it
can
be
made.
A
You
know,
sort
of
seen
in
terms
of
performance
which
is
which
probably
does
need
some
work.
I
don't
know
if
it'll
happen,
but
we
probably
need
to
batch
those
datagrams
in
the
end
when
we
deliver
them
to
the
api
and
maybe
avoid
copying
them
or
something
right
now
it
there's
a
lot
of
cpu
utilization
on
the
when
you're
trying
to
play
a
video
in
the
in
the
demo
browser,
but.
A
Yeah,
I
guess
that's
that's
all
sort
of
dvd
and
and
one
of
the
reasons
I'll
want
to
be
looking
into
the
datagram,
but
I
think
I
think
first
things
once
we
get
to
the
quick,
we'll
be
putting
it
in
the
xhr.
A
I
think
that's
where
it
goes.
I
should
probably
dig
on
that
a
little
further
and
make
sure,
but.
A
D
One
question
I
maybe
had
is
currently
the
api
that
you
know
you
have
in
the
javascript
side
of
things
is
around
the
readable
stream.
As
far
as
I
see,
are
you
also
thinking
about
basically
direct
handover
to
some
of
the
media
devices
api
so
that
you
can,
you
know,
very
easily
paint
to
a
canvas
or
to
a
video
element.
Something
like
that
because
you
seem
to
be
talking
mostly
about
media
related
use
cases.
A
It's
a
good
question
that
was
some.
That
was
another
kind
of
suggestion
in
the
net
dev
chromium
thread
was
that
maybe
we
should
look
into
a
more
specific
use
cases.
D
A
Yeah
and
that's
kind
of
what
we
what
we
said
in
the
in
the
charter,
as
I
recall
that
these
other,
like
doing
something
targeted
to
a
specific
thing,
would
be
fine.
It's
just.
A
So
we
were
aiming
at
web
transport.
First,
we
listed
a
bunch
of
other
apis
that
we
know
might
be
relevant
with
mse.
I
think
as
long
as
you're
as
long
as
you're
building
segments,
then
you
just
pass
them
in.
So
I'm
not
sure
it
needs
to
go
directly
into
msc
itself,
as
opposed
to
generic
transport
of
a
of
an
http
object
in
some
sense.
B
A
A
I
I
would
like
to
avoid
cutting
out
things
like
chat,
for
you
know
for
our
first
for
our
first
cut
stuff,
and
that's
why
I
would
I
would
like
to
still
keep
you
know.
I
think
it
would
be
okay
to
still
make
that
transport
of
http
objects
is
my
sort
of
expectation
for,
for
that
use
case
at
least.
A
And,
and
that
that
that
seems
like,
if
mse
is,
is
accepting
a
sort
of
web
codex
input.
A
Then
yeah
that
might
that
might
make
the
that
I
don't
suppose
you
know
very
much
about
whether
web
codex
can
do
sort
of
unreliable
transport
based
datagrams
like
how
does
it?
How
does
it
handle
that
kind
of
thing?
D
A
Yeah
for
rtp,
I
know
you
get
like
some
sequence,
numbers
and
stuff,
so
you
can
probably
re
put
it
back
in
order
before
you
try
and
handle
it
anyway,
yeah,
as
we
saw
with
kyle.
I
think
it's
it's
just
about
time.
We
are
slightly
over
all
right.
A
Thank
you
for
for
a
good
discussion
and
for
digging
in
on
this
any
parting
words.
Anything
I've
forgotten
that
I
should
know
about
volunteers,
to.
A
To
I
don't
know,
speak
or
or
write
something.