►
From YouTube: 2021-12-01 Multicast W3C Community Group Meeting
Description
Minutes: https://www.w3.org/2021/12/01-multicast-minutes.html
More about the cg: https://github.com/w3c/multicast-cg
A
Okay,
so
we're
recording,
so
I
sent
the
agenda
by
email.
I
did
not
actually
put
together
slides
for
today,
so
I'm
not
going
to
be
sharing
my
screen
or
anything
I'll
just
be
chatting.
But
what
we
have
on
the
agenda
is,
I
wanted
to
discuss
whether
it
would
be
okay
with
those
attending
to
move
this
meeting
back
by
half
an
hour
or
60
minutes.
A
B
B
C
A
A
A
All
right,
so
I
will
tentatively
consider
I'll
I'll
confirm
that
on
the
list
and
see
if
there's
anybody
else
who
raises
objections,
but
I
will
look
to
to
plan
to
move
this
two
hours
later.
Then,
if
that's
that's,
okay,
all
right.
So
the
the
next
item
on
the
list
was
20
minutes
for
a
status
update
last
month
was
reasonably
busy
with
both
the
so
we,
the
last
meeting
we
had
was
was
the
tpack
meeting
that
recording
is
up.
A
A
A
A
My
take
on
this
at
this
stage
is
getting
it
into
the
browser
is
the
step
that
makes
it
most
interesting
to
everybody,
because
then
there's
then
there's
a
distributed
receiver
that
anybody
can
use,
and
so
to
that
end
is
is.
Can
I
I
didn't
hear
anything
that
really
changed
my
position
on
yeah.
We
have
to
get
the
security
stuff
ironed
out
and
we
have
to
get
some
level
of
consensus
built
so
that
the
browsers
would
be
potentially
willing
to
take
it.
A
D
I
think
you're
right
that
getting
it
into
the
browser
is
critical
to
finding
that
break.
Point
and
you're
also
right
that
it's
we're
more
than
more
than
willing,
actually
enthusiastic
about
helping
to
provide
relays
and
as
soon
as
internet2
will
allow
us
to
build
those
overlays
we'll
get
ourselves
back
connected
to
others,
so
that
that
becomes
more
useful
but
yeah.
D
A
A
I
haven't
followed
up
yet,
but
I
will
be
I
mean
I
haven't
followed
up
seriously.
I
did
get.
I
did
ping
the
msec
list
to
verify
that
during
the
session
there
was
some
question:
should
it
be
msec,
or
should
it
be
a
new
mailing
list,
a
pangdem
sect
to
ask?
I
got
unanimous
responses
from
two
people
that
that
msac
seems
like
a
good
place
for
it
for
the
discussion
about
the
security
considerations
for
multicast.
A
A
Not
entirely
unjustified,
but
I
think
that
so
ecker
did
a
review
of
the
draft
and
he
seems
to
think
it's
not
a
good
idea,
but
he
didn't
have
any
technical
objections
that
he
maintained
throughout
during
that
discussion
that
we
had
he.
He
ended
up
in
the
position
that
well
browsers
aren't
interested
in
this.
So
why
should
we
bother
looking
at
it?
A
A
A
A
But
but
it's
I
wouldn't
say
it's
entirely
off
the
table,
but
I'm
gonna
see
if,
if
we
can
do
something
just
with
datagrams
and
web
transport
first,
I
started
setting
up
like
a
local
web
transport,
the
web
platform
tests
to
run
web
transport
against
some
browsers
and
then
I'll
be
hacking
on
that.
I
got
sidetracked
with
ip6
dual
stack
work,
which
of
course
took
much
longer
than
I
was
expecting
so
I'll
speak
about
that
a
little
bit.
B
B
At
what
level
is
it
we're
not
interested
in
this?
Because
we
think
that
the
security
properties
that,
like
that's
too
hard
of
a
problem
to
solve
or
is
it
like,
even
if
we
were
to
develop
a
solution
there,
that
we
think
is
acceptable?
Then
there
then
there
there
are
other
reasons,
perhaps
that
they're
not
yeah.
A
A
A
A
A
A
So
yeah
yeah,
for
example,
the
you
know
when
you
can't
deliver
it
to
a
browser
and
it's
a
it's
a
high
volume
sort
of
event.
What
you
end
up
doing
is
in
order
to
deliver
the
high
volume
event.
You
use
a
a
tv
station
kind
of
approach
right,
you
send
it
to
the
isp
who
sends
it
to
their
to
their
customers
and
now
what
you've
done?
C
A
A
A
Yeah
I
mean
you
know
it's.
It
is
a
concern
that
that
that
people
would
find
out
what
content
is
being
delivered.
That's
one
of
the
ones
that's
been
raised.
Part
of
the
response
is
that
it's
well
a
when
it's
popular
enough,
then
that
that's
very
discoverable
anyway,
through
traffic
analysis.
This
is
even
mentioned
in
the
tls
specs.
A
So
you
have
to
have
everything,
be
it
the
same
size
as
the
biggest
thing
you're
sending
essentially-
or
you
know,
a
large
volume
of
stuff
being
being
at
that
size
and
people
aren't
willing
to
do
that
right.
It's
like
well.
Are
we
gonna
you're,
saying
we're
gonna
triple
the
traffic
that
we're
gonna?
Send
you
know
so
that
we
can
protect
our
users
from
discovering
that
from
their
isp
discovering
if
they
bother
to
look
that
their
that
they're
downloading
call
of
duty,
you
know
what
are
they
gonna
do
with
that
information?
A
A
Like
I,
I
don't,
I
don't
mean
to
diminish
their
concerns,
but
but
I
think
that
most
of
the
concerns
that
people
have
are
coming
from
a
place
of
not
having
yet
engaged
with
the
problem
seriously
and
with
the
problems
they're
raising
and
what
the
actual
impact
on
users
is,
and
the
scalability
you
know
is
just
it's
something.
That's
hard
to
turn
your
back
on
once
you
understand
how
big
it
is.
A
C
D
A
That
gets
this
great
pet
dns
data,
not
not
everyone
on
the
path,
and
so
there's
there's
sort
of
a
an
analogy
here
where
there
is
a
bunch
of
multicast
traffic
happening
today.
Nobody
knows:
what's
what
people
are
analyzing
it
and
what
they're
getting
out
of
it?
And
so
you
know
if
there
is
really
a
privacy
problem
here,
then
it's
a
privacy
problem,
that's
present
today
and
and
it's
something
that
that
we
should
be
thinking
about
and
discussing.
A
A
C
A
A
A
A
And
those
I'm
bringing
them
up
and
down
still
there's
still
some
pieces
that
I
haven't
been
able
to
to
properly
test
or
get
working,
but
the
relays
I'm
trying
to
get
functioning
and
then
my
intent
is
to
so
giant
also
has
brought
up
some
some
multicast
capability
in
their
multicast
working
group.
They
actually
have
like
a
beer
forward,
plane
and
and
some
pretty
neat
relays
running.
A
They
gave
me
an
idea
because
they're
playing
some
some
public
domain
music
over
their
multicast
channel,
which
you
can
just
leave
running
you
don't
have
to
be
watching
it
right.
So
I
I
might
do
the
same
for
some
of
my
sources
with
the
through
vlc,
where
you
can
just
leave
it
running
and
listen
to
some
music.
A
A
A
A
This
challenge
is
at
the
aws
level,
just
getting
that
dual
stack
functioning,
but
that
part
is
the
part,
that's
working,
but
then
inside
the
device.
That's
the
that's
running
there.
It's
not
behaving
the
same.
So
if
this
is
a
an
ubuntu
sender
in
ip4,
what
I'm
doing
is
I
create
a
v
pair
and-
and
I
just
set
the
route
to
send
it
to
one
side
of
the
v
pair.
A
Unfortunately,
the
with
ip6
for
reasons
I've
not
yet
been
able
to
track
down
setting
the
route
does
not
seem
to
be
sufficient.
It's
going
out
the
the
the
physical
interface
whether
or
not
I
have
a
route
to
go
to
the
thief
pair
interface,
and
I'm
I'm
waffling
between
setting
this
up
inside
a
net
namespace
versus.
So
I
know
from
previously
working
on
it
that
if
I
set
it
up
as
a
separate
device,
then
the
multicast
traffic
will
not
transfer
between
the
the
two
aws
instances.
A
Into
the
relay
somehow
or
into
some
other
kind
of
tunnel,
I
don't
know
yet
why
the
routing
isn't
behaving
the
way
I
expect
it
to,
and
I
think
the
network
name
space
is
the
most
promising
way
to
to
force
it.
I
thought
about
trying
it
in
a
verf.
Also
I
I
I
I
might,
but
that
seems
like
it
has
challenges
too.
Anyway,
I've
been
on
this
for
a
couple
of
weeks
now,
I'm
hoping
to
actually
end
up
with
it
running,
but.
A
It's
it's
not
it's
not
what
I
wish
it
was
for
sure.
So
I
guess
anyone
who
who
wants
to
work
on
that
with
me.
Please
do
let
me
know
and
I'll
I'll
hook
you
up
and
see.
If
you
can
help
me
troubleshoot
it
but
yeah,
it's
it's
not
fantastic,
but
that
one
when
it
when
I
do
set
it
up
does
have
the
same
sort
of
I.
I
did
get
the
relay
discovery.
A
Part
working
for
that,
so
that
doing
the
same
dryad
lookup,
with
the
with
the
reverse
ip
of
the
of
the
source
address,
does
find
the
the
relay
that
it's
supposed
to
be
running
on.
I
just
don't
have
actual
traffic
sending
there,
so
I
might
give
this
another
week
before
I
abandon
it,
but
I'm
hopeful
that
in
a
week
I'll
I'll
get
through
all
the
hoops
that
need
to
happen,
but
I
really
do
want
to
get
onto
the
quick
bit.
A
A
A
A
A
D
A
This
is
basically
the
the
proof
of
concept
work
that
we
did
with
the
with
the
four
with
the
four
other
isps
they
set
this
up
in
lab.
There
was
one
that
did
it
across
their
production
network,
but
but
the
receiver
was
still
lab
set
up
and
yeah
I
mean,
if
you're,
if
you're
willing
to
do
that,
then
I
think
that
that
could
be.
That
could
be
good
and
we
can.
We
can
bring
up
the
the
relay
connectivity.
B
Yeah,
so
a
question
I
I
wasn't
able
to
join
the
second
screen
group
meeting
did:
was
there
anything
useful
that
came
out
of
that?
I'm
just
aware
that
that
group
was
has
has
been
designing
protocols
for
use
like
on
a
local
network
that
you
know
have
web
compatible
security
properties,
so
it
may
be
wonderful.
A
A
A
I
think
they
were
more
interested
in
that
were
like
as
tls
connections
to
local
devices
that
ended
up
having
some
some
discussion
on
github.
You
know,
I
think
a
few
people
that
were
in
that
meeting
came
to
the
tpac
discussion
for
multicast,
but
they
did
not
later
join
the
group
or
attend
this
meeting.
Unfortunately,
so
so
I
don't
think
I
sufficiently
interested
them
to
to
actually
get
involved
at
this
stage.
So.
A
You
know
to
say
what
specific
kinds
of
services
or
names
you're
going
to
be
trying
to
discover
with
your
application
and
then
and
then
apple
has
some
degree
of
control
over
over
the
apps
that
are
allowed
to
do
that.
Apparently,
they
saw
some
abuses
from
from
some
sdks,
often
that
the
apps
didn't
even
know
were
were
doing
sketchy
things,
but
the
sdks
were
were
probing
the
local
network
in
some
ways,
and
so
they
so
they
lock
down
a
number
of
local
discovery.
Sorts
of
things
you
can
take
a
look
at.
A
A
A
A
You
know,
there's
a
principle
of
data
minimization
or
something
that
you
know
just
want
to
expose
all
this
stuff
yeah.
They
seemed
to.
They
seemed
like
they're
in
a
reasonably
similar
place.
There's
like
they
haven't,
really
got
much
of
an
answer.
They
don't
really
know
what
the
threat
model
is
that
they
need
to
protect
against.
A
You
know,
yeah.
I
didn't
see
anything
that
was
super
helpful.
I
was
hoping
they
had
already
just
sort
of
solved
this
problem,
but
I
think
it's
some
indication
that
the
local
network
part
is
not
necessarily
what
people
are
worried
about
at
least
right
now,
although
I'm
not
sure
it's
what
they
shouldn't
it's
a
thing
they
shouldn't
be
worried
about,
but
in
terms
of
browser
security
and
the
feedback
that
people
have
gotten
yet
yeah.
A
So
far,
I
think
it's
been
driven
more
by
responses
to
to
abuses,
but
we're
seeing
people
push
back
on
things
that
open
up
just
new
information.
That's
exposed,
because
every
new
piece
of
information
that's
exposed
has
some
kind
of
fingerprinting
at
least
and
maybe
starts
to
to
present
new
threat
models
that
we
don't
know
what
they
are,
and
so
in
that
sense
it
just
sort
of
continues
raising
the
bar
for
for
what
all
needs
to
happen
in
terms
of
answering
all
the
security
questions.